ErrLookup › jax-ml/jax
jax-ml/jax
Composable transformations of Python+NumPy programs: differentiate, vectorize, JIT to GPU/TPU, and more · Python · 867 source files
Analyzed at 1e1c6a8fc0 on 2026-08-27. 2300 documented errors.
| Code / Message | Type | Severity | Tags |
|---|---|---|---|
| numpy masked arrays are not supported as direct inputs to JA | validation | error | pytree, custom-node, jax, tree-flatten, api-contract |
| Python int {value} too large to convert to int64 | validation | error | pytree, custom-node, jax, tree-flatten-with-path |
| Python int {value} too large to convert to int32 | validation | error | pytree, namedtuple, jax, tree-flatten |
| The names should be exclusive and should not intersect in `n | validation | error | pytree, namedtuple, jax, pickle, tree-flatten-with-path |
| The return value of the policies should be a boolean. Got: { | validation | warning | pytree, deprecation, sets, generators, jax |
| prevent_cse must be a bool or tuple of bools, got {type(prev | validation | error | pytree, unflatten, jax, leaf-count |
| the `static_argnums` argument to `jax.checkpoint` / `jax.rem | validation | error | pytree, unflatten, jax, leaf-count |
| the `static_argnums` argument to `jax.checkpoint` / `jax.rem | validation | error | pytree, tree-map, structure-mismatch, jax |
| Effects not supported in AD of `checkpoint`/`remat`: {disall | error_code | error | pytree, none-handling, jax, tree-map, breaking-change |
| non-bool prevent_cse together with static_argnums/static_arg | error_code | error | pytree, jax, structure-mismatch, from-iterable |
| the rematted computation's closure contains a mutable array | validation | error | pytree, jax, registry, compose |
| Effects not supported in partial-eval of `checkpoint`/`remat | error_code | error | pytree, jax, registry, tuple |
| stop_gradient only works on valid JAX arrays, but input argu | validation | error | pytree, jax, pickle, version-mismatch |
| {self.__class__.__name__} has no attribute {name} | error_code | error | pytree, jax, pickle, malformed-data |
| Mapped away dimension of inputs passed to vmap should be sha | validation | error | jax, vmap, sharding, mesh, distributed |
| Unmapped values passed to vmap cannot be sharded along the m | validation | error | jax, vmap, sharding, mesh |
| {name} wrapped function must be passed at least one argument | validation | error | jax, vmap, axis-size, api-misuse |
| {name} was requested to map a value of non-array type {core. | validation | error | jax, vmap, in-axes, type-error |
| {name} was requested to map its argument along axis {axis}, | validation | error | jax, vmap, shape, rank, in-axes |
| {name} must have at least one non-None value in in_axes or a | validation | error | jax, vmap, in-axes, axis-size |
| {''.join(msg)[:-2]} | validation | error | jax, vmap, shape, batch-size |
| primal and tangent arguments to jax.jvp must be tuples or li | validation | error | jax, jvp, autodiff, api-misuse, type-error |
| primal and tangent arguments to jax.jvp must have the same t | validation | error | jax, jvp, pytree, autodiff |
| primal and tangent arguments to jax.jvp do not match; dtypes | validation | error | jax, jvp, dtype, autodiff, float0 |
| jvp called with different primal and tangent shapes;Got prim | validation | error | jax, jvp, shape, autodiff |
| expected {in_avals.tree}, got {tangents_ft.tree} | validation | error | jax, linearize, jvp, pytree |
| linearized function called on tangent values inconsistent wi | validation | error | jax, linearize, tangent, aval, mixed-precision |
| reduce_axes argument to vjp is deprecated | validation | error | jax, vjp, deprecated, migration |
| unexpected JAX type (e.g. shape/dtype) for gradient ref pass | validation | error | jax, vjp, refs, gradient-accumulation, dtype |
| the gradient for {_vjp_arg_name(jaxpr, in_tree, idx)}, which | validation | error | jax, vjp, refs, autodiff |
| {_vjp_arg_name(jaxpr, in_tree, idx)} is Ref-typed, so its gr | validation | error | jax, vjp, refs, autodiff |
| But the tree structures differ: | validation | error | jax, vjp, pytree, refs |
| the VJP function was applied before restoring its not-saveab | validation | error | jax, vjp, memory, checkpointing, saveable-args |
| Accumulator aval mismatch: expected {aval}, got {acc.aval} | validation | error | jax, vjp, accumulator, internal-api, dtype |
| {full_name} must be a pytree prefix with bool leaves or a tu | validation | error | jax, pytree, validation, bool-prefix |
| {full_name} must form a tree prefix of the corresponding val | validation | error | jax, pytree, tree-prefix |
| {full_name} must form a tree prefix of the corresponding val | validation | error | jax, pytree, arity |
| structure of the differentiated function {jaxpr.debug_info.f | validation | error | jax, vjp, autodiff, pytree |
| unexpected JAX type (e.g. shape/dtype) for argument to VJP f | validation | error | jax, vjp, dtype, shape-mismatch |
| {_vjp_too_many_args(name, len(extra_args) + 1)} | validation | error | jax, vjp, arity, argument-count |
| reduce_axes argument to transpose is deprecated | validation | error | jax, linear-transpose, deprecated-api |
| linear_transpose only supports [float or complex] -> [float | validation | error | jax, linear-transpose, dtype |
| cotangent tree does not match function output, expected {out | validation | error | jax, linear-transpose, pytree |
| cotangent type does not match function output, expected {out | validation | error | jax, linear-transpose, dtype, shape-mismatch |
| iteration over a 0-d array | exception | error | jax, numpy-interop, iteration, scalar, typeerror |
| to_dlpack can only pack a dlpack tensor from an array on a s | exception | error | jax, dlpack, interop, sharding, multi-device |
| __dlpack__ only supported for unsharded arrays. | exception | error | jax, dlpack, sharding, buffer-protocol, interop |
| Unknown GPU platform for __dlpack__: {platform_version} | exception | error | jax, dlpack, gpu, platform-detection, version-mismatch |
| Couldn't get local_hardware_id for __dlpack__ | exception | error | jax, dlpack, gpu, device-init, environment |
| __dlpack__ device only supported for TPU pinned host memory | exception | error | jax, dlpack, tpu, pinned-memory, interop |
| __dlpack__ device only supported for CPU, GPU and TPU pinned | exception | error | jax, dlpack, backend, unsupported-platform, interop |
| unsafe_buffer_pointer() is supported only for unsharded arra | exception | error | jax, cuda, raw-pointer, sharding, unsupported-operation |
| __cuda_array_interface__() is supported only for unsharded a | exception | error | jax, cuda-array-interface, cupy, sharding, interop |
| Buffer callbacks do not support transpose. Please use `jax.c | exception | error | jax, autodiff, gradient, callback, buffer-callback |
| multi-platform lowering for buffer_callback | exception | error | jax, xla, multi-platform, lowering, export |
| `buffer_callback` not supported on {platform} backend. | exception | error | jax, backend, tpu, unsupported-platform, buffer-callback |
| buffer_callback callback must not return any values. | exception | error | jax, callback, host-callback, return-value, buffer-callback |
| jax.pure_callback failed to find a local CPU device to place | exception | error | jax, environment, pure-callback, cpu-device, jax-platforms |
| Pure callbacks do not support JVP. Please use `jax.custom_jv | exception | error | jax, autodiff, jvp, pure-callback, custom-jvp |
| Pure callbacks do not support transpose. Please use `jax.cus | exception | error | jax, autodiff, vjp, transpose, pure-callback, custom-vjp |
| callbacks are only supported in spmd computations when all m | exception | error | jax, spmd, sharding, mesh, pure-callback, manual-axes |
| callbacks do not support specifying sharding inside spmd com | exception | error | jax, spmd, sharding, pure-callback, manual-axes |
| pure_callback only supports SingleDeviceSharding, but got {t | exception | error | jax, sharding, pure-callback, single-device-sharding |
| Partitioned callback not implemented on {platform} backend. | exception | error | jax, callback, tpu, sharding, backend-support |
| Partitioned callback not supported with return values. | exception | error | jax, callback, sharding, api-contract |
| Mismatched number of outputs from callback. Expected: {}, Ac | exception | error | jax, callback, runtime-validation, shape-mismatch |
| Incorrect output shape for return value #{i}: Expected: {out | exception | error | jax, callback, shape-mismatch, runtime-validation |
| Incorrect output dtype for return value #{i}: Expected: {out | exception | error | jax, callback, dtype-mismatch, runtime-validation |
| {str(exc)} | exception | error | jax, checkify, assertion, debugging |
| Checkify does not support batched while-loops (checkify-of-v | exception | error | jax, checkify, vmap, while-loop, transformation-order |
| Mesh must be provided for shard_map with checkify. | exception | error | jax, checkify, shard-map, mesh, sharding |
| Unsupported aval type: {type(v)} | exception | error | jax, checkify, shard-map, custom-aval, extensibility |
| {prim_name} takes a scalar pred as argument, got {pred} | exception | error | jax, checkify, type-error, scalar-predicate |
| Formatting arguments to checkify.check need to be PyTrees of | exception | error | jax, checkify, typeerror, assertions, jit |
| check_error takes an Error as argument, got type {type(error | exception | error | jax, checkify, typeerror, api-misuse |
| `compute_on`'s compute_type argument must be a string. | exception | error | jax, type-error, compute-on, argument-validation |
| Nesting `compute_on` with different compute types is not all | validation | error | jax, compute-on, nesting, lowering |
| invalid truth value {val!r} for environment {varname!r} | validation | error | jax, environment-variable, config, bool-parsing |
| Unrecognized config option: {name} | exception | error | jax, config, unknown-option, typo |
| For flags with a corresponding contextmanager, read their va | exception | error | jax, config, contextmanager, api-migration |
| Config option {name} already defined | exception | error | jax, config, duplicate-registration, startup |
| bool() not supported for instances of type '{0}' (did you me | exception | error | jax, config, typeerror, truthiness |
| Context manager for {state.__name__} config option requires | exception | error | jax, config, contextmanager, missing-argument |
| Default value must be of type bool, got {default} of type {g | exception | error | jax, config, typeerror, bool-default |
| Default value must be of type str, got {default} of type {ge | exception | error | jax, config, typeerror, enum |
| Invalid value "{default}" for JAX flag {name} | validation | error | jax, config, valueerror, environment-variable, enum |
| new enum value must be in {enum_values}, got {new_val} of ty | validation | error | jax, config, valueerror, enum |
| Default value must be of type str or None, got {default} of | exception | error | jax, config, typeerror, enum |
| new enum value must be None or in {enum_values}, got {new_va | validation | error | jax, config, valueerror, enum |
| Default value must be of type {enum_class}, got {default} of | exception | error | jax, config, typeerror, enum |
| Invalid value "{default_str}" for JAX flag {name} | validation | error | jax, config, valueerror, environment-variable, enum |
| new enum value must be an instance of {enum_class}, got {new | validation | error | jax, config, typeerror, enum |
| Default value must be of type int, got {default} of type {ge | exception | error | jax, config, typeerror, int |
| Invalid value "{default_env}" for JAX flag {name} | validation | error | jax, config, valueerror, environment-variable, int |
| new int config value must be None or of type int, got {new_v | validation | error | jax, config, valueerror, int |
| Default value must be of type float, got {default} of type { | exception | error | jax, config, typeerror, float |
| new float config value must be None or of type float, got {n | validation | error | jax, config, valueerror, float |
| new string config value must be of type str, got {new_val} o | validation | error | jax, config, typeerror, string |
| new string config value must be None or of type str, got {ne | validation | error | jax, config, valueerror, string |
| Value of type {type(self)} is not convertible to float. | exception | error | jax, tracer, type-conversion, jit, float |
| Value of type {type(self)} is not convertible to complex. | exception | error | jax, tracer, complex-numbers, jit |
| Value of type {type(self)} is not convertible to hex. | exception | error | jax, tracer, hex, formatting, debugging |
| Value of type {type(self)} is not convertible to oct. | exception | error | jax, tracer, oct, formatting |
| Value of type {type(self)} is not convertible to integer ind | exception | error | jax, tracer, indexing, operator-index, control-flow |
| The error occurred in the __reduce__ method, which may indic | exception | error | jax, tracer, pickle, serialization, multiprocessing, caching |
| Value of type {type(self)} is not indexable. | exception | error | jax, tracer, immutable-arrays, setitem, functional-updates |
| Value of type {type(self)} is not compatible with the Array | exception | error | jax, tracer, numpy-interop, array-api, nep-35 |
| The '{name}' method is not available on {self._error_repr()} | exception | error | jax, tracer, block-until-ready, async, attribute-access |
| The 'sharding' attribute is not available on {self._error_re | exception | error | jax, tracer, sharding, attribute-access, distributed |
| {self.__class__.__name__} has no attribute {name} | exception | error | jax, tracer, attributeerror, jit, duck-typing |
| The addressable_data() method was called on {self._error_rep | exception | error | jax, concretization, tracer, sharding, addressable-data |
| The delete() method was called on {self._error_repr()}.{self | exception | error | jax, concretization, tracer, memory-management, delete |
| The devices() method was called on {self._error_repr()}.{sel | exception | error | jax, concretization, tracer, devices, placement |
| The global_shards property was called on {self._error_repr() | exception | error | jax, concretization, tracer, sharding, global-shards |
| The is_deleted() method was called on {self._error_repr()}.{ | exception | error | jax, concretization, tracer, memory-management, is-deleted |
| The is_fully_addressable property was called on {self._error | exception | error | jax, concretization, tracer, sharding, distributed |
| The is_fully_replicated property was called on {self._error_ | exception | error | jax, concretization, tracer, sharding, replication |
| The on_device_size_in_bytes() method was called on {self._er | exception | error | jax, concretization, tracer, memory, profiling |
| The traceback property was called on {self._error_repr()}.{s | exception | error | jax, concretization, tracer, traceback, debugging |
| The unsafe_buffer_pointer() method was called on {self._erro | exception | error | jax, tracing, tracer, jit, buffer-pointer |
| unbound axis name: {axis_name} | exception | error | jax, axis-name, vmap, pmap, collectives |
| Triggering __jax_array__() during abstractification is no lo | exception | error | jax, pytree, jax-array, custom-class, tracing |
| Cannot interpret value of type {typ} as an abstract array; i | exception | error | jax, type-error, abstract-value, tracing, dtype |
| Argument '{x}' of type '{typ}' is not a valid JAX type | exception | error | jax, type-error, jit, argument-validation |
| TracerBoolConversionError | exception | error | jax, tracer, bool-conversion, control-flow, jit |
| TracerIntegerConversionError | exception | error | jax, tracer, integer-conversion, indexing, jit |
| The problem arose with the `{fname}` function. | exception | error | jax, tracer, concretization, jit, float-conversion |
| Sharding spec {spec} implies that array axis {dim} is partit | exception | error | jax, sharding, named-sharding, mesh, partitionspec |
| Length of sharding.spec ({len(out_s.spec)}) must be equal to | exception | error | jax, sharding, partition-spec, shape-mismatch |
| Mesh of an aval must be an AbstractMesh. Got {out_s.mesh} of | exception | error | jax, sharding, mesh, abstract-mesh, type-mismatch |
| sharding with memory_kind is not allowed. Please use `jax.de | exception | error | jax, sharding, memory-kind, device-put |
| Axes mentioned in `manual_axis_type` field of ShapedArray sh | exception | error | jax, sharding, manual-axis-type, mesh, validation |
| varying and unreduced cannot have common mesh axes. Got vary | validation | error | jax, sharding, manual-axis-type, validation |
| varying and reduced cannot have common mesh axes. Got varyin | validation | error | jax, sharding, manual-axis-type, validation |
| Expected unreduced_kind to be of type `jax.sharding.Unreduce | validation | error | jax, sharding, unreduced-kind, type-error |
| `unreduced_kind` should be `None` when `unreduced` is an emp | validation | error | jax, sharding, unreduced-kind, validation |
| {name} argument of ManualAxisType should of type `frozenset` | validation | error | jax, sharding, type-error, constructor-validation |
| The layout of ShapedArray should not be `AutoLayout` when la | validation | error | jax, layout, auto-layout, configuration |
| len() of unsized object | exception | error | jax, numpy, len, zero-d-array |
| {name} cannot accept args which are unreduced. Got {a.str_sh | validation | error | jax, sharding, collectives, unreduced, spmd |
| {name} cannot accept args with unreduced_kind={a.mat.unreduc | validation | error | jax, sharding, collectives, unreduced-kind, spmd |
| {name} cannot accept args which are reduced. Got {a.str_shor | validation | error | jax, sharding, collectives, reduced, spmd |
| Primitive {prim_name} requires varying manual axes to match, | validation | error | jax, shard-map, spmd, internal-error, workaround |
| array ref with memory space only works inside of a `jit`. | exception | error | jax, refs, memory-space, jit, not-implemented |
| pinned array ref only works inside of a `jit`. | exception | error | jax, refs, pinned-memory, jit, not-implemented |
| {self!r} has no `sharding`. | exception | error | jax, aval, attribute-error, tokens |
| Unsupported sequence length Q {T}, KV {S}. | exception | error | jax, cudnn, flash-attention, sequence-length, training |
| Packed layout requires a GPU with at least Hopper architectu | exception | error | jax, cudnn, gpu-architecture, packed-attention, hopper |
| mla requires cudnn version >= 9.10 and at least hopper arch. | exception | error | jax, cudnn, mla, version-mismatch, gpu-architecture |
| cuDNN is not detected. | exception | critical | jax, cudnn, installation, environment, missing-dependency |
| Currently only support batch_dim in [0, None], but got {dim= | exception | error | jax, vmap, cudnn, batching, in-axes |
| Query, key and value should have same sharding. | validation | error | jax, sharding, gspmd, multi-device, cudnn |
| Sharding on sequence dim is not allowed. | validation | error | jax, sharding, sequence-parallel, spmd, cudnn |
| Sharding on head dim is not allowed. | validation | error | jax, sharding, tensor-parallel, spmd, cudnn |
| Query and bias should have same sharding on batch and num_he | validation | error | jax, sharding, attention-bias, spmd, cudnn |
| Sharding on bias sequence dim is not allowed. | validation | error | jax, sharding, attention-bias, sequence-parallel, cudnn |
| Require q_seqlen and kv_seqlen to use packed layout | validation | error | jax, attention, varlen, packed-layout, cudnn |
| shardings should container 4 inputs, but got {len(shardings) | validation | error | jax, fp8, sharding, spmd, matmul |
| shardings specs rank should be 3, but got lhs: {len(lhs.spec | validation | error | jax, fp8, sharding, rank-mismatch, matmul |
| shardings spec for batch dim should be same, but got lhs: {l | validation | error | jax, fp8, sharding, batch-dim, matmul |
| Unrecognized mode: {config.mode}. | validation | error | jax, fp8, quantization, config, cudnn |
| Only support preferred_element_type in (f32, bf16, f16), but | validation | error | jax, dtype, fp8, matmul, unsupported-type |
| The input arguments to the custom_vmap-decorated function {d | validation | error | jax, custom-vmap, kwargs, argument-binding |
| No batching rule defined for custom_vmap function {debug_fun | exception | error | jax, custom-vmap, vmap, missing-rule |
| structure of output value and output batching specification | validation | error | jax, custom-vmap, pytree, vmap |
| structure of output returned by custom vmap rule ({rule_name | validation | error | jax, custom-vmap, pytree, vmap |
| Can't use ``defjvps`` with ``nondiff_argnums``. | validation | error | jax, custom-jvp, autodiff, api-misuse |
| No JVP defined for custom_jvp function {primal_name} using d | exception | error | jax, custom-jvp, autodiff, missing-rule |
| The input arguments to the custom_jvp-decorated function {pr | validation | error | jax, custom-jvp, kwargs, argument-binding |
| Custom JVP rule {jvp_name} for function {primal_name} must p | validation | error | jax, custom-jvp, autodiff, return-shape |
| Custom JVP rule {jvp_name} for function {primal_name} must p | validation | error | jax, custom-jvp, pytree, autodiff |
| Custom JVP rule {jvp_name} for function {primal_name} must p | validation | error | jax, custom-jvp, pytree, autodiff |
| Custom JVP rule {jvp_name} for function {primal_name} must p | validation | error | jax, custom-jvp, shape, dtype, autodiff |
| Custom JVP rule must produce primal and tangent outputs with | validation | error | jax, custom-jvp, tangent, shape, dtype |
| Custom JVP rule must produce primal and tangent outputs with | validation | error | jax, custom-jvp, tangent, pytree, autodiff |
| Effects not supported in `custom_jvp`: {disallowed_effects} | exception | error | jax, custom-jvp, effects, purity, jit |
| custom_gradient function used with with_logs=True must retur | exception | error | jax, custom-gradient, typeerror, vjp |
| The inputs to the closure produced by closure_convert must h | exception | error | jax, closure-convert, pytree, typeerror |
| transpose output pytree structure must match that of linear | exception | error | jax, linear-call, transpose, pytree |
| remat optimization for custom_vjp does not support forward f | exception | error | jax, custom-vjp, remat, side-effects |
| remat optimization for custom_vjp does not support higher-or | exception | error | jax, custom-vjp, remat, higher-order-ad |
| keyword arguments could not be resolved to positions | exception | error | jax, custom-partitioning, kwargs, typeerror |
| Custom Partitioning rules must return Sharding. | exception | error | jax, custom-partitioning, sharding, tpu |
| Mismatch in result shapes. %s vs %s | exception | error | jax, custom-partitioning, sharding, shape-mismatch |
| Unknown keyword arguments: {sharding_rule_dict} | exception | error | jax, custom-partitioning, shardy, sharding-rule |
| Shardy is used, but sharding propagation callbacks instead o | exception | error | jax, shardy, custom-partitioning, migration, tpu |
| Custom-partitioned function {function!r} does not support GS | exception | error | jax, gspmd, shardy, custom-partitioning, config |
| sharding_rule callable must produce either an SdyShardingRul | exception | error | jax, shardy, sharding-rule, custom-partitioning |
| Factor names have to start with a letter, but got '{factor[0 | exception | error | jax, shardy, sharding-rule, validation |
| Unknown character '{char}' | exception | error | jax, shardy, sharding-rule, validation |
| A compound factor should contain at least two factors | exception | error | jax, shardy, sharding-rule, validation |
| Each element of CompoundFactor must be a str, but got {type( | exception | error | jax, shardy, sharding-rule, typeerror |
| Ellipsis can't be used in a compound factor | exception | error | jax, shardy, sharding-rule, ellipsis |
| Each element of ArrayMapping must be a str or CompoundFactor | exception | error | jax, shardy, sharding-rule, typeerror |
| Ellipsis can only be used at the beginning of a dimension | exception | error | jax, shardy, sharding-rule, ellipsis |
| Factor {factor} is not used in the rule, but size is provide | exception | error | jax, sharding, factor-sizes, validation |
| Factor {factor} is only used in compound factors; must speci | exception | error | jax, sharding, compound-factor, factor-sizes |
| Factor {factor} represents a whole dimension; do not specify | exception | error | jax, sharding, factor-sizes, redundant-config |
| {kind} must be a tuple of factors | exception | error | jax, sharding, type-error, tuple |
| {kind} contains duplicated factors | validation | error | jax, sharding, duplicate-values, validation |
| Factor {factor} in {kind} is not used in the rule | validation | error | jax, sharding, unknown-factor, validation |
| Factor {factor} can only be in one of the reduction, need re | validation | error | jax, sharding, conflicting-config, validation |
| Compound factors should be one level, nested brackets are no | validation | error | jax, sharding, compound-factor, rule-syntax |
| Brackets are not balanced | validation | error | jax, sharding, unbalanced-parens, rule-syntax |
| Brackets should contain at least two factors | validation | error | jax, sharding, compound-factor, rule-syntax |
| Factor names have to start with a letter, but got '{char}' | validation | error | jax, sharding, naming, rule-syntax |
| Brackets are not balanced in rule: '{rule}' | validation | error | jax, sharding, unbalanced-parens, rule-syntax |
| rule must be a str, but got {type(rule)} | validation | error | jax, sharding, type-error, rule-syntax |
| factor_sizes must be a dict of str to int, but got {factor_s | validation | error | jax, sharding, type-error, factor-sizes |
| Unknown character '{BATCHING}' | validation | error | jax, sharding, reserved-character, rule-syntax |
| Character '.' must be used inside ellipsis '...' | validation | error | jax, sharding, ellipsis, rule-syntax |
| There is no -> in rule: '{rule}' | validation | error | jax, sharding, missing-separator, rule-syntax |
| Sharding rule has {len(rule.operand_mappings)} operands, but | validation | error | jax, sharding, custom-partitioning, mlir, sdy |
| Argument '{arg}' of type {type(arg)} is not a valid JAX type | validation | error | jax, type-validation, argument-validation |
| nan | exception | error | jax, nan, numerical-debugging |
| inf | exception | error | jax, inf, numerical-overflow |
| For a cross-host reshard in multi-controller JAX, input and | validation | error | jax, multi-controller, resharding, device-mismatch |
| mTLS for the JAX distributed service requires jaxlib 0.11.2 | exception | error | jax, distributed, mtls, version-mismatch |
| coordinator_address should be defined. | validation | error | jax, distributed, coordinator, config |
| Number of processes must be defined. | validation | error | jax, distributed, world-size, config |
| The process id of the current process must be defined. | validation | error | jax, distributed, rank, config |
| process_id must be a nonnegative int. Got process_id={proces | validation | error | jax, distributed, type-error, rank |
| num_processes must be a positive int. Got num_processes={num | validation | error | jax, distributed, type-error, world-size |
| process_id and num_processes must be nonnegative, with proce | validation | error | jax, distributed, rank-validation |
| coordinator_bind_address should be defined. | validation | error | jax, distributed, network, bind-address |
| distributed.initialize should only be called once. | exception | error | jax, distributed, double-init, lifecycle |
| Preemption sync manager should only be initialized once. | exception | error | jax, distributed, preemption, singleton |
| jax.distributed.initialize() must be called before any JAX c | exception | critical | jax, distributed, xla-backend, ordering |
| Expected a string or dtype-like object; got {dtype=} | validation | error | jax, dlpack, dtype, none-guard |
| Specified {device=} which requires a copy since the source d | validation | error | jax, dlpack, device-mismatch, copy |
| Argument to to_dlpack must be a jax.Array, got {type(x)} | validation | error | jax, dlpack, type-check, typeguard |
| The device specification passed to to_dlpack contains an uns | exception | error | jax, dlpack, device-type, buffer-protocol |
| JAX does not support any version below {MIN_DLPACK_VERSION} | exception | error | jax, dlpack, version-negotiation, protocol |
| Specified {device=} which requires a copy since the source d | validation | error | jax, dlpack, device-mismatch, copy |
| from_dlpack can only unpack a dlpack tensor onto a singular | validation | error | jax, dlpack, sharding, multi-device |
| The array passed to from_dlpack must have __dlpack__ and __d | validation | error | jax, dlpack, type-check, buffer-protocol |
| Array passed to from_dlpack is on unsupported device type (D | validation | error | jax, dlpack, device-type, interop |
| Specified input which requires a copy since the source data | validation | error | jax, dlpack, alignment, zero-copy |
| JAX requires ml_dtypes version 0.5 or newer; installed versi | validation | critical | jax, dependency, version, ml-dtypes |
| dtype cannot be None. | validation | error | jax, dtype, none-guard, validation |
| unexpected input: {dtype=} | validation | error | jax, dtype, validation |
| dtype {dtype!r} not understood | validation | error | jax, dtype, type-check, canonicalization |
| Invalid scalar value {x} | validation | error | jax, scalar, type-check |
| Python int {value} too large to convert to {dtype} | validation | error | jax, overflow, int64, x64 |
| Unrecognized {kind=} expected one of {list(_dtype_kinds.keys | validation | warning | jax, dtype, array-api, validation |
| Expected kind to be a dtype, string, or tuple; got {kind=} | validation | error | jax, dtype, isdtype, argument-validation |
| {dtype=} is not a valid dtype for JAX type promotion. | validation | error | jax, type-promotion, dtype, lattice |
| Input dtypes {tuple(str(n) for n in nodes)} have no availabl | validation | error | jax, type-promotion, casting, dtype |
| Dtype {dtype} is not a valid JAX array type. Only arrays of | validation | error | jax, dtype, validation, numeric-types |
| Explicitly requested dtype {}{} is not available. To enable | validation | error | jax, x64, dtype, configuration |
| Invalid argument to dtype: {x}. | validation | error | jax, dtype, none-argument |
| Value '{x}' with dtype {dt} is not a valid JAX array type. O | validation | error | jax, dtype, string-dtype, validation |
| Cannot determine dtype of {x} | validation | error | jax, dtype, array-like, inference-failure |
| at least one array or dtype is required | validation | error | jax, result-type, argument-validation |
| dtype must be specified. | validation | error | jax, dtype, argument-validation |
| Passing an array as a dtype argument is no longer supported; | validation | error | jax, dtype, api-removal, migration |
| JAX only supports number, bool, and string dtypes, got dtype | validation | error | jax, dtype, validation, numeric-types |
| iteration over a 0-d array | validation | error | jax, earray, iteration, zero-d-array |
| len() of unsized object | validation | error | jax, earray, len, zero-d-array |
| The error code state and the predicate must be on the same m | validation | error | jax, error-checking, sharding, mesh, distributed |
| raise_if_error() should not be called within a traced contex | validation | error | jax, error-checking, jit, tracer |
| No VJP is available | validation | error | jax, export, vjp, serialization |
| Duplicate serialization registration for type `{nodetype}`. | validation | error | jax, export, serialization, registry, duplicate-registration |
| Duplicate serialization registration for serialized_name `{s | validation | error | jax, export, serialization, registry, name-collision |
| serializing DisabledSafetyCheck: {check} | exception | error | jax, export, serialization, flatbuffers |
| Cannot deserialize DisabledSafetyCheck with unknown kind: {k | exception | error | jax, export, deserialization, version-mismatch |
| Encountered dimension variable '{self.var}' that is not appe | exception | error | jax, export, shape-polymorphism, dimension-variables |
| Cannot divide {self} by {divisor}. | exception | error | jax, shape-polymorphism, floor-division, export |
| __pow__ modulo not implemented | exception | error | jax, shape-polymorphism, pow, not-implemented |
| Symbolic dimension cannot be raised to non-integer powers: ' | exception | error | jax, shape-polymorphism, pow, validation |
| Symbolic dimension cannot be raised to negative powers: '{se | exception | error | jax, shape-polymorphism, pow, negative-exponent |
| __rpow__ modulo not implemented | exception | error | jax, shape-polymorphism, rpow, not-implemented |
| Symbolic dimension '{self}' used in a context that requires | exception | error | jax, shape-polymorphism, int-conversion, tracing |
| The symbolic constraints should be a sequence of strings. Go | exception | error | jax, shape-polymorphism, constraints, api-misuse |
| SymbolicScope constraint must be a string: got {repr(c_str)} | exception | error | jax, shape-polymorphism, constraints, type-validation |
| Constraint parsing error: must contain one of '==' or '>=' o | exception | error | jax, shape-polymorphism, constraints, parsing |
| Unsatisfiable explicit constraint: {constr.debug_str} | exception | error | jax, shape-polymorphism, constraints, unsatisfiable |
| Invalid equality constraint: {e1} == {e2}. The left-hand-sid | exception | error | jax, shape-polymorphism, constraints, normalization |
| Found multiple equality constraints with the same left-hand- | exception | error | jax, shape-polymorphism, constraints, duplicate-key |
| Invalid mixing of symbolic scopes {when}.\nExpected {self_de | exception | error | jax, shape-polymorphism, scope, export |
| Symbolic dimension comparison {cmp_str()} is inconclusive.{d | exception | error | jax, shape-polymorphism, comparison, inconclusive |
| Symbolic dimension {operation_name} not supported for {p}. | exception | error | jax, shape-polymorphism, type-error, coercion |
| Encountered unexpected shape dimension {d} | exception | error | jax, export, shape-polymorphism, type-error |
| for grad support, subclass {type(self)} must implement `vjp_ | validation | error | jax, autodiff, custom-primitive, not-implemented |
| for jvp support, subclass {type(self)} must implement `jvp` | validation | error | jax, autodiff, forward-mode, not-implemented |
| for linearize support, subclass {type(self)} must implement | validation | error | jax, autodiff, linearize, not-implemented |
| for transpose support, subclass {type(self)} must implement | validation | error | jax, autodiff, transpose, not-implemented |
| for vmap support, subclass {type(self)} must implement `batc | validation | error | jax, vmap, batching, custom-primitive |
| under vmap, the {member_name} of {name} produced an output b | validation | error | jax, vmap, custom-vjp, custom-jvp, batching |
| input type mismatch for {_prim} | validation | error | jax, type-mismatch, tracing, custom-primitive |
| {type(_prim).__name__} returned structured residuals from `v | validation | error | jax, autodiff, vjp, api-contract |
| tree mismatch during linearization of {prim=}. Expected: {pr | error_code | error | jax, pytree, linearize, autodiff |
| {type(_prim).__name__}.vjp_bwd should return None or a dict | validation | error | jax, autodiff, vjp, api-contract |
| {type(_prim).__name__}.transpose should return None or a dic | validation | error | jax, autodiff, transpose, api-contract |
| subclass {type(self)} can't set both `jvp = jvp_from_lin` an | validation | error | jax, autodiff, circular-definition, misconfiguration |
| `linearize_from_jvp` is a pair of rules, not a single rule; | validation | error | jax, api-misuse, unpacking |
| `vjp_from_jvp` is a pair of rules, not a single rule; unpack | validation | error | jax, api-misuse, unpacking |
| `vjp_from_lin` is a pair of rules, not a single rule; unpack | validation | error | jax, api-misuse, unpacking |
| Custom VJP bwd rule {self.bwd} was registered with defvjp_wi | exception | error | jax, custom-vjp, api-contract, logging |
| Custom VJP bwd rule {self.bwd} was registered with defvjp_wi | exception | error | jax, custom-vjp, logging, type-validation |
| Custom VJP bwd rule {self.bwd} must produce a tuple but got | exception | error | jax, custom-vjp, return-type |
| Custom VJP bwd rule {self.bwd} must produce a tuple of lengt | exception | error | jax, custom-vjp, arity-mismatch |
| Effects not supported in `custom_jvp`: {disallowed} | exception | error | jax, custom-jvp, autodiff, effects |
| at {keystr(path)}, got fwd output type {ty.str_short()} whic | exception | error | jax, custom-vjp, autodiff, type-mismatch |
| {result}the bwd rule attached to {primal_sourceinfo} produce | exception | error | jax, custom-vjp, autodiff, type-mismatch |
| No VJP defined for custom_vjp function {self.f.__name__} usi | exception | error | jax, custom-vjp, missing-rule, autodiff |
| custom_vjp inputs marked with nondiff_argnums must be static | exception | error | jax, custom-vjp, tracer-error, nondiff-argnums |
| Custom JVP rule {jvp_name} for function {self.traced.fun_nam | exception | error | jax, custom-jvp, autodiff, return-contract |
| Custom JVP rule {jvp_name} for function {self.traced.fun_nam | exception | error | jax, custom-jvp, pytree-mismatch |
| Custom JVP rule {jvp_name} for function {self.traced.fun_nam | exception | error | jax, custom-jvp, pytree-mismatch |
| Custom JVP rule must produce primal and tangent outputs with | exception | error | jax, custom-jvp, tangent-mismatch |
| Custom JVP rule must produce primal and tangent outputs with | exception | error | jax, custom-jvp, tangent-mismatch, multi-output |
| Can't use ``defjvps`` with ``nondiff_argnums``. | exception | error | jax, custom-jvp, defjvps, api-misuse |
| No JVP defined for custom_jvp function {self.f.__name__} usi | exception | error | jax, custom-jvp, missing-rule, autodiff |
| The input arguments to the custom_jvp-decorated function {se | exception | error | jax, custom-jvp, kwargs, signature-binding |
| custom_jvp inputs marked with nondiff_argnums must be static | exception | error | jax, custom-jvp, tracer-error, nondiff-argnums |
| custom_jvp-decorated function {self.f} closed over a {type(t | exception | error | jax, custom-jvp, tracer-error, closure |
| Sharded hypothesis runner does not support `data()` inside ` | exception | error | jax, testing, hypothesis, sharding |
| Unknown resize method "{s}" | validation | error | jax, image-resize, invalid-argument |
| shape must have length equal to the number of dimensions of | validation | error | jax, image-resize, shape-mismatch |
| No constant handler for type: {type(val)} | exception | error | jax, constants, type-error, tracing |
| numpy masked arrays are not supported as direct inputs to JA | exception | error | numpy, jax, masked-array, input-validation |
| A ShapeDtypeStruct does not have a value and cannot be used | exception | error | jax, shape-dtype-struct, constants |
| Unsupported scalar attribute type: {type(val)} | exception | error | jax, dtype, mlir, numpy |
| NumPy arrays with zero strides are not supported as MLIR att | exception | error | numpy, strides, jax, mlir |
| No attribute handler defined for type: {type(val)} | exception | error | jax, mlir, custom-primitives, attributes |
| accessing .backend in multi-lowering setting. This can occur | exception | error | jax, lowering, multi-platform, primitive |
| the platform for the specified backend {xb.canonicalize_plat | exception | error | jax, platform-mismatch, lowering, backend |
| Host callback lowering created too many channels. PjRt does | exception | error | jax, host-callback, resource-limit, debugging |
| Invalid compute type {c_type}. Current supported values are | exception | error | jax, mlir, input-validation, enum, compute-type |
| Cannot lower jaxpr with effects: {closed_jaxpr.effects} | exception | error | jax, effects, custom-partitioning, lowering, pmap |
| custom_call backend_config unexpected type: {backend_config} | exception | error | jax, mlir, custom-call, backend-config, type-error |
| Error refining shapes. {dump_module_message(module, "before_ | exception | error | jax, polymorphic-shapes, dynamic-shapes, mlir, lowering |
| Cannot use `partial_eval_jaxpr_custom` with stateful jaxprs. | validation | error | jax, remat, partial-eval, stateful-jaxpr, control-flow |
| Value returned by a remat policy should be a bool or `ad_che | validation | error | jax, remat, checkpoint-policy, type-error, ad-checkpoint |
| `JaxprInputEffect` {eff} does not have a corresponding jaxpr | validation | error | jax, jaxpr, effects, internal-invariant, mutable-arrays |
| {primitive}.abstract_eval() method should return a tuple or | validation | error | jax, custom-primitives, abstract-eval, multiple-results, api-contract |
| re-tracing function {fun} for `jit`, but 'no_tracing' is set | exception | error | jax, config, no-tracing, cache-miss, aot |
| function {dbg.func_src_info} traced for {dbg.traced_for} ret | validation | error | jax, tracing, type-error, return-value, jit |
| function {dbg.func_src_info} traced for {dbg.traced_for} ret | validation | error | jax, mutable-arrays, refs, tracing, return-value |
| numpy masked arrays are not supported as direct inputs to JA | validation | error | jax, numpy, masked-array, input-validation, sharding |
| No pxla_result_handler for type: {type(aval)} | error_code | error | jax, pxla, custom-aval, result-handler, extension-api |
| JAX tried to execute function {self.name}, but the no_execut | exception | error | jax, config, no-execution, environment, diagnostics |
| unbound axis name: {axis_name} | validation | error | jax, mesh, collectives, axis-name, sharding |
| AbstractMesh should be of the same size across all shardings | validation | error | jax, sharding, abstract-mesh, pjit, mesh-size-mismatch |
| AbstractMesh size: {abstract_mesh.size} does not match the d | validation | error | jax, sharding, abstract-mesh, device-assignment, pjit |
| The following ordered effects are not supported for more tha | validation | error | jax, effects, multi-device, ordered-effects, pjit, debug-print |
| Passing lowering_platforms via jax.export or jit(f).trace(*a | validation | error | jax, abstract-mesh, lowering-platforms, export, aot, pjit |
| State effect not supported in vmap-of-cond. | exception | error | jax, vmap, cond, state-effects, experimental |
| IO effect not supported in vmap-of-cond. | exception | error | jax, vmap, cond, io-callback, host-callback |
| State effect not supported in cond partial-eval. | exception | error | jax, cond, partial-eval, state-effects, autodiff |
| lax.platform_dependent: the '{pname}' branch must be a calla | validation | error | jax, lax, platform-dependent, typeerror, callable |
| Use 'cuda', 'rocm', or 'oneapi' for lax.platform_dependent. | validation | error | jax, lax, platform-dependent, gpu, invalid-argument |
| lax.platform_dependent: the 'default' branch must be a calla | validation | error | jax, lax, platform-dependent, typeerror, callable |
| Effects not supported in `scan`: {disallowed_effects} | exception | error | jax, scan, effects, scan3, experimental |
| `unroll` must be a `bool` or a non-negative `int`. | validation | error | jax, scan, unroll, valueerror, argument-validation |
| lax.scan: f argument should be a callable. | validation | error | jax, scan, typeerror, callable, api-misuse |
| zero-length scan is not supported in disable_jit() mode beca | validation | error | jax, scan, disable-jit, zero-length, edge-case |
| scan body output must be a pair, got {}. | validation | error | jax, scan, typeerror, return-shape, pytree |
| must provide `length` to `scan`, since the leading-axis size | validation | error | jax, scan, hijax, length, experimental |
| scan got value with no leading axis to scan over: {}. | validation | error | jax, scan, shape-inference, scalar-operand, pytree |
| 0th dimension of all xs should be replicated. Got {} | validation | error | jax, scan, sharding, multi-device, jaxmesh |
| scan got `length` argument of {} which disagrees with leadin | validation | error | jax, scan, length-mismatch, shape-validation |
| scan got values with different leading axis sizes: {}. | validation | error | jax, scan, shape-mismatch, length-inference |
| scan got no values to scan over and `length` not provided. | validation | error | jax, scan, missing-length, argument-validation |
| {} function carry input and carry output must have the same | validation | error | jax, pytree, scan, while-loop, carry-structure |
| {} function carry input and carry output must have equal typ | validation | error | jax, dtype-mismatch, scan, while-loop, shard-map, vma |
| scan number of arguments doesn't match the number of jaxpr a | validation | error | jax, internal, scan, jaxpr, argument-count |
| Scan carry input and output got mismatched varying manual ax | validation | error | jax, shard-map, vma, scan, spmd, multi-device |
| lax.while_loop: body_fun and cond_fun arguments should be ca | validation | error | jax, while-loop, typeerror, callable, api-misuse |
| cond_fun must return a boolean scalar, but got pytree {}. | validation | error | jax, while-loop, cond-fun, pytree, boolean-scalar |
| cond_fun must return a boolean scalar, but got output type(s | validation | error | jax, while-loop, cond-fun, dtype, scalar |
| Effects not supported in `while`: {} | error_code | error | jax, effects, while-loop, debug-print, not-implemented |
| Ordered IO effects not supported in vmap. | error_code | error | jax, vmap, ordered-io, debug-print, while-loop, batching |
| Unordered IO effects not supported in while_loop with batche | error_code | error | jax, vmap, while-loop, io-effects, control-flow |
| Reverse-mode differentiation does not work for lax.while_loo | validation | error | jax, autodiff, grad, while-loop, fori-loop, dynamic-bounds |
| Cannot write to the same ref in both cond and body of while | error_code | error | jax, state-api, refs, while-loop, state-discharge |
| Body jaxpr has consts. If you see this error, please open an | error_code | error | jax, state-discharge, internal-error, while-loop |
| Cond jaxpr has consts. If you see this error, please open an | error_code | error | jax, state-discharge, internal-error, while-loop |
| lax.fori_loop: body_fun argument should be callable. | validation | error | jax, fori-loop, typeerror, argument-validation |
| lower and upper arguments to fori_loop must have equal types | validation | error | jax, fori-loop, dtype-mismatch, typeerror |
| Can only use `unroll` in `fori_loop` if the loop bounds are | validation | error | jax, fori-loop, unroll, static-shape, traced-values |
| 0th dimension of leaf passed to `jax.lax.map` should be repl | validation | error | jax, lax-map, sharding, multi-device, gspmd |
| lax.associative_scan: fn argument should be callable. | validation | error | jax, associative-scan, type-error, callable |
| associative scan over axis of non-constant size: {}. You may | error_code | error | jax, associative-scan, dynamic-shape, shape-polymorphism |
| Array inputs to associative_scan must have the same first di | validation | error | jax, associative-scan, shape-mismatch |
| XLA operations do not allow negative axes | validation | error | jax, lax, cumsum, negative-axis |
| axis {} is out of bounds for array of shape {} | validation | error | jax, lax, cumsum, axis-out-of-bounds |
| {} does not accept dtype {}. Accepted dtypes are subtypes of | validation | error | jax, lax, dtype, cumsum |
| {}() output shapes must match {}, got {} and {} | validation | error | jax, custom-linear-solve, shape-mismatch, linear-algebra |
| transpose_solve required for backwards mode automatic differ | validation | error | jax, autodiff, custom-linear-solve, transpose |
| open an issue at https://github.com/google/jax !! | error_code | error | jax, autodiff, custom-linear-solve, not-implemented |
| String padding is not implemented for transposed convolution | validation | error | jax, lax, convolution, transposed-conv, padding |
| padding argument to conv_general_dilated should be a string | validation | error | jax, lax, convolution, padding |
| Invalid padding mode: {padding} | validation | error | jax, lax, conv-transpose, padding |
| No 4+ dimensional dimension_number defaults. | validation | error | jax, lax, conv-transpose, dimension-numbers, high-rank |
| conv_general_dilated lhs and rhs must have the same number o | validation | error | jax, lax, convolution, shape-mismatch |
| conv_general_dilated feature_group_count must be a positive | validation | error | jax, lax, convolution, feature-group-count |
| conv_general_dilated feature_group_count must divide lhs fea | validation | error | jax, lax, convolution, feature-group-count |
| conv_general_dilated lhs feature dimension size divided by f | validation | error | jax, lax, convolution, feature-group-count, kernel-shape |
| conv_general_dilated rhs output feature dimension size must | validation | error | jax, lax, convolution, feature-group-count, kernel-shape |
| conv_general_dilated batch_group_count must be a positive in | validation | error | jax, lax, convolution, batch-group-count |
| conv_general_dilated batch_group_count must divide lhs batch | validation | error | jax, convolution, shape-validation, grouped-conv |
| conv_general_dilated rhs output feature dimension size must | validation | error | jax, convolution, shape-validation, grouped-conv |
| At most one of batch_group_count and feature_group_count may | validation | error | jax, convolution, grouped-conv, api-misuse |
| conv_general_dilated window and window_strides must have the | validation | error | jax, convolution, strides, shape-validation |
| Convolutions with non-static strides, dilation, feature_grou | exception | error | jax, jit, static-shapes, convolution, lowering |
| Wrong number of explicit pads for convolution: expected {}, | validation | error | jax, convolution, padding, shape-validation |
| Negative padding is larger than the size of the correspondin | validation | error | jax, convolution, padding, shape-validation |
| convolution requires lhs and rhs ndim to be equal, got {} an | validation | error | jax, convolution, shape-validation, rank-mismatch |
| convolution dimension_numbers list/tuple must be length 3, g | validation | error | jax, convolution, dimension-numbers, api-misuse |
| convolution dimension_numbers elements must be strings, got | validation | error | jax, convolution, dimension-numbers, type-error |
| convolution dimension_numbers[{}] must have len equal to the | validation | error | jax, convolution, dimension-numbers, shape-validation |
| convolution dimension_numbers must be tuple/list or None, go | validation | error | jax, convolution, dimension-numbers, type-error |
| convolution dimension_numbers[{}] must contain the character | validation | error | jax, convolution, dimension-numbers, validation |
| convolution dimension_numbers[{}] cannot have duplicate char | validation | error | jax, convolution, dimension-numbers, validation |
| convolution dimension_numbers elements must each have the sa | validation | error | jax, convolution, dimension-numbers, validation |
| Unknown FFT type '{s}' | validation | error | jax, fft, api-misuse, invalid-enum-value |
| Unknown FFT type value '{fft_type}' | validation | error | jax, fft, type-error |
| only real valued inputs supported for rfft | validation | error | jax, fft, dtype, api-misuse |
| full must be called with scalar fill_value, got fill_value.s | validation | error | jax, lax, type-error, scalar-required, shape-validation |
| offset must be an integer, got {offset!r} | validation | error | jax, lax, tri, integer-required, type-error |
| repeated axis in lax.expand_dims: {dimensions} | validation | error | jax, lax, expand-dims, duplicate-axis, value-error |
| repeated axis in lax.expand_dims: {dims} | validation | error | jax, lax, expand-dims, axis-canonicalization, duplicate-axis |
| Invalid dimension range passed to collapse: {operand.shape}[ | validation | error | jax, lax, collapse, dimension-range, value-error |
| Arguments to batch_matmul must be at least 2D, got {}, {} | validation | error | jax, lax, batch-matmul, dimensionality, value-error |
| Arguments to batch_matmul must have same ndim, got {}, {} | validation | error | jax, lax, batch-matmul, rank-mismatch, value-error |
| Called {name} with a float0 array. float0s do not support an | validation | error | jax, float0, autodiff, gradient, type-error |
| {} does not accept dtype {}. Accepted dtypes are subtypes of | exception | error | jax, lax, dtype-validation, type-error |
| {name} does not accept dtype {dtype_to_string(aval.dtype)}. | exception | error | jax, lax, narrow-int, int4, dtype-validation |
| unreduced rule for {name} is not implemented. Please file an | exception | error | jax, sharding, gspmd, not-implemented, unreduced |
| Called {name} with a float0 at position {i}. float0s do not | exception | error | jax, float0, autodiff, binop, type-error |
| {} does not accept dtype {} at position {}. Accepted dtypes | exception | error | jax, lax, binop, dtype-validation, type-error |
| {}: arrays must have same number of dimensions, got {}. | exception | error | jax, sharding, named-sharding, rank-mismatch, type-error |
| transpose with implicit broadcasting of unshaped values. Got | exception | error | jax, autodiff, vjp, dynamic-shapes, transpose, type-error |
| the first argument to pow must have an inexact dtype (float | exception | error | jax, lax, pow, dtype-mismatch, type-error |
| Integers cannot be raised to negative powers, got integer_po | exception | error | jax, lax, integer-pow, negative-exponent, type-error |
| The denominator cannot be unreduced passed to `div`. Got {y= | exception | error | jax, sharding, div, unreduced, value-error |
| Denominator should be reduced along the same axes numerator | exception | error | jax, sharding, lax, division, distributed |
| HLO comparison {direction} for extended dtype {avals_in[0].d | exception | error | jax, extended-dtype, comparison, lax, hlo |
| `sharding` passed to `convert_element_type` can only contain | exception | error | jax, sharding, dtype-conversion, lax |
| can only convert to an extended dtype on an array type,but g | exception | error | jax, extended-dtype, abstract-eval, dynamic-shape, typeerror |
| Cannot convert_element_type from {dtype_to_string(x.dtype)} | exception | error | jax, extended-dtype, dtype-conversion, validation |
| can only convert to extended dtype from its representation d | exception | error | jax, extended-dtype, dtype-conversion, representation |
| can only convert to extended dtype from an array of its repr | exception | error | jax, extended-dtype, shape-validation, rank-mismatch |
| can only convert to extended dtype from an array of its repr | exception | error | jax, extended-dtype, shape-validation, suffix-mismatch |
| can only convert to extended dtype from an array with traili | exception | error | jax, extended-dtype, sharding, partition-spec |
| can only convert from an extended dtype on an array type,but | exception | error | jax, extended-dtype, abstract-eval, typeerror |
| Cannot convert_element_type from {dtype_to_string(x.dtype)} | exception | error | jax, extended-dtype, dtype-conversion, validation |
| can only convert from extended dtype to its representation d | exception | error | jax, extended-dtype, dtype-conversion, representation |
| Attempting to convert array of shape {operand.shape} from {o | exception | error | jax, bitcast, dtype, shape-validation |
| lax.bitcast_convert_type does not support bool or complex va | exception | error | jax, bitcast, complex, bool, typeerror |
| Input type is incompatible with `preferred_element_type`. Th | exception | error | jax, dot-general, preferred-element-type, dtype-validation |
| `preferred_element_type` must have the same signedness as th | exception | error | jax, dot-general, signedness, dtype-validation |
| `preferred_element_type` must not be narrower than the origi | exception | error | jax, dot-general, precision, dtype-validation |
| dot_general requires lhs dimension numbers to be nonnegative | exception | error | jax, dot-general, dimension-numbers, index-out-of-range |
| dot_general requires rhs dimension numbers to be nonnegative | exception | error | jax, dot-general, dimension-numbers, index-out-of-range |
| dot_general requires equal numbers of lhs_batch and rhs_batc | exception | error | jax, dot-general, batch-dims, dimension-numbers |
| dot_general requires lhs batch dimensions to be distinct, go | exception | error | jax, dot-general, dimension-numbers, batch-dims |
| dot_general requires rhs batch dimensions to be distinct, go | exception | error | jax, dot-general, dimension-numbers, batch-dims |
| dot_general requires lhs contracting dimensions to be distin | exception | error | jax, dot-general, dimension-numbers, contraction |
| dot_general requires rhs contracting dimensions to be distin | exception | error | jax, dot-general, dimension-numbers, contraction |
| dot_general requires lhs batch dimensions to be disjoint fro | exception | error | jax, dot-general, dimension-numbers, batch-dims, contraction |
| dot_general requires rhs batch dimensions to be disjoint fro | validation | error | jax, dot-general, dimension-numbers, batch-dims, contraction |
| dot_general requires lhs batch dimensions and rhs batch dime | validation | error | jax, dot-general, shape-mismatch, batch-dims |
| dot_general requires contracting dimensions to have the same | validation | error | jax, dot-general, shape-mismatch, contraction |
| `out_sharding` passed to `dot_general` can only contain unre | validation | error | jax, dot-general, sharding, spmd, experimental-api |
| {name} argument type error: {lhs.dtype}, {rhs.dtype} | validation | error | jax, dot-general, dtype-mismatch, mixed-precision |
| `preferred_element_type` must not be narrower than the origi | validation | error | jax, dot-general, preferred-element-type, dtype-narrowing |
| Accuracy {accuracy} not supported | exception | error | jax, dot-general, accuracy, experimental-api |
| The precision '{precision}' is not supported by dot_general | validation | error | jax, dot-general, cpu-backend, precision, fp8 |
| lhs_ragged_dim {lhs_ragged_dim} not found in lhs_noncontract | validation | error | jax, ragged-dot-general, dimension-numbers |
| ragged_dot_general requires {dim_name} numbers to be nonnega | validation | error | jax, ragged-dot-general, index-out-of-range |
| ragged_dot_general expects exactly one lhs ragged dimension. | validation | error | jax, ragged-dot-general, dimension-numbers |
| expected rank of group_sizes to be >=1. | validation | error | jax, ragged-dot-general, group-sizes, shape-mismatch |
| expected group_sizes to have shape {expected_gs_shape}, got | validation | error | jax, ragged-dot-general, group-sizes, shape-mismatch |
| ragged_dot_general requires the group count (last dimension | validation | error | jax, ragged-dot-general, dynamic-shapes, jit |
| ragged_dot_general requires zero group dimensions in the rhs | validation | error | jax, ragged-dot, shape-validation, moe |
| ragged_dot_general requires exactly one rhs group dimension | validation | error | jax, ragged-dot, shape-validation, moe |
| ragged_dot_general requires rhs group dimension numbers to b | validation | error | jax, ragged-dot, dimension-numbers, shape-validation |
| expected rhs group dimension size to be {num_groups}, got {r | validation | error | jax, ragged-dot, shape-mismatch, moe |
| ragged_dot_general requires that group_sizes.dtype is subtyp | validation | error | jax, ragged-dot, dtype-validation |
| Unimplemented group_offset support. | exception | error | jax, ragged-dot, autodiff, not-implemented |
| ragged_dot vmap over any dim but 0 - NYI | exception | error | jax, ragged-dot, vmap, batching, not-implemented |
| Explicit sharding inference for ragged_dot_general is not cu | exception | error | jax, ragged-dot, sharding, multi-device, spmd |
| broadcast_in_dim broadcast_dimensions must have length equal | validation | error | jax, broadcast-in-dim, shape-validation |
| broadcast_in_dim target broadcast shape must have equal or h | validation | error | jax, broadcast-in-dim, rank-validation |
| broadcast_in_dim broadcast_dimensions must be a subset of ou | validation | error | jax, broadcast-in-dim, index-out-of-range |
| broadcast_in_dim operand dimension sizes must either be 1, o | validation | error | jax, broadcast-in-dim, shape-mismatch |
| broadcast_in_dim broadcast_dimensions must not contain dupli | validation | error | jax, broadcast-in-dim, duplicate-index |
| `out_sharding` passed to `broadcast_in_dim` can only contain | validation | error | jax, broadcast-in-dim, sharding, gspmd, unreduced |
| reps length must be equal to the ndim of x, got {len(reps)=} | validation | error | jax, tile, shape-validation |
| clamp requires min.shape == operand.shape or min.shape == () | validation | error | jax, clamp, shape-mismatch, broadcasting |
| clamp requires max.shape == operand.shape or max.shape == () | validation | error | jax, clamp, shape-mismatch, broadcasting |
| concatenate expects at least one operand, got 0. | validation | error | jax, concatenate, shape-validation, empty-input |
| All objects to concatenate must be arrays, got {}. | validation | error | jax, concatenate, type-validation, dtype |
| Cannot concatenate arrays with different numbers of dimensio | validation | error | jax, concatenate, rank-mismatch, shape-validation |
| concatenate dimension out of bounds: dimension {} for shapes | validation | error | jax, concatenate, axis-out-of-bounds, shape-validation |
| Cannot concatenate arrays with shapes that differ in dimensi | validation | error | jax, concatenate, shape-mismatch, shape-validation |
| stack expects at least one operand, got 0. | validation | error | jax, stack, empty-input, shape-validation |
| Cannot stack arrays with different numbers of dimensions: go | validation | error | jax, stack, rank-mismatch, shape-validation |
| All input arrays must have the same shape. Got {}. | validation | error | jax, stack, shape-mismatch |
| unstack requires arrays with rank > 0, however a scalar arra | validation | error | jax, unstack, scalar, rank-error |
| Sizes passed to split must be nonnegative, got {list(sizes)} | validation | error | jax, split, negative-size, validation |
| Sum of sizes {np.sum(sizes)} must be equal to dimension {axi | validation | error | jax, split, size-mismatch, shape-validation |
| pad operand and padding_value must be same dtype: got {} and | validation | error | jax, pad, dtype-mismatch |
| padding_value must be a scalar; got {np.shape(padding_value) | validation | error | jax, pad, scalar-required, shape-validation |
| length of padding_config must equal the number of axes of op | validation | error | jax, pad, config-mismatch, rank-error |
| interior padding in padding_config must be nonnegative, got | validation | error | jax, pad, negative-padding, validation |
| Dimension size after padding is not at least 0, got result s | validation | error | jax, pad, negative-dimension, shape-validation |
| dimensions are not unique: {dimensions} | validation | error | jax, squeeze, duplicate-dims, validation |
| dimensions outside range [0, ndim): {dimensions} | validation | error | jax, squeeze, axis-out-of-bounds |
| cannot select an axis to squeeze out which has size not equa | validation | error | jax, squeeze, size-one-required, shape-validation |
| reshape new_sizes must all be positive, got {}. | validation | error | jax, reshape, negative-dimension, shape-validation |
| duplicate value in 'axes' of reduction: {axes} | validation | error | jax, lax, reduction, axes, duplicate |
| reduction axes {axes} contains out-of-bounds indices for {op | validation | error | jax, lax, reduction, out-of-bounds, axis |
| out_sharding passed to {name} cannot be reduced. Got {out_sh | validation | error | jax, sharding, named-sharding, reduction, distributed |
| reduce_sum on operand {operand.str_short(True)} is not allow | validation | error | jax, float16, bfloat16, reduction, mixed-precision, config-flag |
| Invalid axis {axis} for operand shape {operand.shape} | validation | error | jax, argmax, argmin, axis, validation |
| argmin and argmax require non-empty reduced dimension. opera | validation | error | jax, argmax, empty-array, shape-validation |
| index_dtype must be an integer type, but got {} | validation | error | jax, argmin, dtype, type-error |
| logical reduction requires operand dtype bool or int, got {o | validation | error | jax, lax, logical-reduction, dtype |
| reduce_precision: exponent_bits must be positive; got {expon | validation | error | jax, reduce-precision, bit-manipulation, validation |
| reduce_precision: mantissa_bits must be non-negative; got {m | validation | error | jax, reduce-precision, quantization, validation |
| Arguments to sort must have equal shapes, got: {shapes} | validation | error | jax, lax, sort, shape-mismatch |
| top_k is not compatible with complex inputs. | validation | error | jax, top-k, complex-numbers, dtype |
| top_k operand must have >= 1 dimension, got {} | validation | error | jax, top-k, scalar, rank-error |
| axis argument out of range: {axis=} for {operand.shape=} | validation | error | jax, top-k, axis, out-of-bounds |
| k argument to top_k must be no larger than size along axis; | validation | error | jax, top-k, shape-validation, off-by-one |
| top_k returns int32 indices, which will overflow for array d | validation | error | jax, top-k, int32-overflow, large-arrays |
| Arguments to after_all must be tokens | validation | error | jax, lax, tokens, control-flow, xla |
| Arguments to rng_uniform must have identical dtypes, got {} | validation | error | jax, rng-uniform, dtype-mismatch |
| Arguments to rng_uniform must be scalars; got shapes {} and | validation | error | jax, rng-uniform, scalar-requirement, shape-validation |
| Unsupported dtype: {dtype} | validation | error | jax, linalg, eig, dtype, gpu |
| Nonsymmetric eigendecomposition requires cusolver 11.7.1 or | error_code | error | jax, cuda, cusolver, eig, version-mismatch, gpu |
| Left eigenvectors are not supported by cusolver | validation | error | jax, eig, cusolver, gpu, not-implemented |
| Derivatives of non-symmetric eigenvectors are only valid und | validation | error | jax, autodiff, eig, gradients, numerical |
| Argument to symmetric eigendecomposition must have shape [.. | validation | error | jax, linalg, eigh, shape-validation, square-matrix |
| subset_by_index not supported on CPU and GPU | validation | error | jax, eigh, subset-by-index, cpu, gpu, not-implemented |
| QDWH implementation is only supported on TPU | validation | error | jax, eigh, qdwh, tpu, backend, not-implemented |
| Jacobi implementation is not supported on CPU | validation | error | jax, eigh, jacobi, cpu, backend, not-implemented |
| Derivatives not defined for partial eigen decomposition. | validation | error | jax, autodiff, eigh, subset-by-index, gradients |
| Argument to Hessenberg reduction must have shape [..., n, n] | validation | error | jax, linalg, hessenberg, shape-validation |
| hessenberg requires the last dimension of a to be constant, | validation | error | jax, hessenberg, dynamic-shapes, jit, cpu |
| The first argument to householder_product must have at least | validation | error | jax, linalg, householder, shape-validation, qr |
| The second argument to householder_product must not have mor | validation | error | jax, linalg, householder, shape-validation |
| ormqr with left=True expects c to have the same number of ro | validation | error | jax, linalg, ormqr, shape-validation, qr |
| ormqr with left=False expects c to have the same number of c | validation | error | jax, linalg, ormqr, shape-validation, qr |
| 'trans' value must be 0, 1, or 2, got {trans} | validation | error | jax, lu-solve, invalid-argument, api-misuse |
| last two dimensions of LU decomposition must be equal, got s | validation | error | jax, lu-solve, shape-validation, square-matrix |
| b matrix must have rank >= 1, got shape {b.shape} | validation | error | jax, lu-solve, shape-validation, rank |
| When LU decomposition matrix and b have the same number of d | validation | error | jax, lu-solve, shape-validation, broadcasting |
| When LU decomposition matrix and b different numbers of dime | validation | error | jax, lu-solve, shape-validation, broadcasting |
| logaddexp2 requires floating-point or complex inputs; got {x | validation | error | jax, dtype, lax, validation |
| This gmm kernel only supports either (m, k) x (g, k, n) -> ( | validation | error | jax, pallas, gpu, shape, moe |
| Group sizes {group_sizes.shape=} must match first dimension | exception | error | jax, pallas, moe, shape, gpu |
| Explicit sharding is not currently supported in the pallas-t | exception | error | jax, sharding, pallas, spmd, not-implemented |
| group_offset is not currently supported in the pallas-triton | exception | error | jax, pallas, gpu, not-implemented, ragged-dot |
| {name} cannot accept args with unreduced_kind={mat.unreduced | exception | error | jax, shard-map, spmd, collectives, internal-api |
| axis_index_groups only supported for sums over just named ax | exception | error | jax, psum, axis, validation |
| pargmin only accepts a single axis, got {axis_name} | exception | error | jax, pargmin, axis, typeerror |
| axis_index_groups must cover all indices exactly once | exception | error | jax, psum, validation, axis-index-groups |
| `perm` passed to `jax.lax.ppermute` must be a list or a tupl | exception | error | jax, ppermute, typeerror, collectives |
| `perm` does not represent a permutation: {perm} | exception | error | jax, pshuffle, permutation, validation |
| The size of all_to_all split_axis ({x.shape[split_axis]}) ha | exception | error | jax, all-to-all, shape, spmd |
| all_to_all requires the size of the mapped axis axis_name to | exception | error | jax, all-to-all, shape, spmd |
| axis_index_groups not supported in vmap collectives. Please | exception | error | jax, vmap, collectives, not-implemented |
| axis_index_groups can only be used with reductions over name | exception | error | jax, collectives, axis-validation |
| Found an unbound axis name: {name}. To fix this, please call | exception | error | jax, mesh, shard-map, unbound-axis |
| axis_index_groups must all be the same size for TPU lowering | exception | error | jax, tpu, collectives, lowering |
| {op_name} sources and destinations must be unique, got {}. | exception | error | jax, ppermute, collectives, validation |
| Make sure that the axis_name passed to jax.lax.ppermute is i | exception | error | jax, ppermute, mesh, ordering |
| psend is currently only implemented on GPUs | exception | error | jax, psend, gpu-only, backend |
| psend currently only supports manual sharding | exception | error | jax, psend, shard-map, manual-sharding |
| psend is currently only implemented on GPU | exception | error | jax, psend, backend, not-implemented |
| precv currently only supports manual sharding | exception | error | jax, precv, shard-map, manual-sharding |
| precv is currently only implemented on GPU | exception | error | jax, precv, backend, not-implemented |
| pbroadcast batcher only supports a single axis | exception | error | jax, pbroadcast, vmap, not-implemented |
| Replica groups must be equally sized | exception | error | jax, all-to-all, collectives, validation |
| Please open a feature request! | exception | error | jax, all-to-all, vmap, not-implemented |
| all_to_all must be used within a mapped context like vmap or | exception | error | jax, all-to-all, mapped-context |
| ragged_all_to_all input_offsets must be integer type. | exception | error | jax, ragged-all-to-all, dtype |
| ragged_all_to_all send_sizes must be integer type. | exception | error | jax, ragged-all-to-all, dtype |
| ragged_all_to_all output_offsets must be integer type. | exception | error | jax, ragged-all-to-all, dtype |
| ragged_all_to_all recv_sizes must be integer type. | exception | error | jax, ragged-all-to-all, dtype |
| pvary is a invariant->variant collective. This means that th | validation | error | jax, collectives, mesh, spmd, pvary |
| all_gather_reduced only accepts inputs that are varying. Got | validation | error | jax, collectives, mesh, spmd, all-gather |
| all_gather_reduced is a Varying -> Reduced collective. This | validation | error | jax, collectives, mesh, axis-name, all-gather |
| all_gather_reduced's input cannot be reduced across the axis | validation | error | jax, collectives, mesh, reduced, all-gather |
| Please file an issue at https://github.com/jax-ml/jax/issues | exception | error | jax, vmap, collectives, not-implemented, all-gather |
| unreduced_psum_scatter only accepts inputs that are unreduce | validation | error | jax, collectives, mesh, spmd, reduce-scatter |
| unreduced_psum_scatter is a Unreduced -> Varying collective. | validation | error | jax, collectives, axis-name, mesh, reduce-scatter |
| unreduced_psum_scatter's input cannot be varying across the | validation | error | jax, collectives, mesh, varying, reduce-scatter |
| {name} only accepts inputs that are unreduced. Got {aval.str | exception | error | jax, sharding, collectives, validation, mesh |
| {name} is a Unreduced -> Invariant collective. This means th | exception | error | jax, mesh, axis-name, collectives, sharding |
| {name}'s input cannot be varying across the axis_name provi | exception | error | jax, sharding, collectives, varying, mesh |
| {name} does not accept integer axis_name. Got axis_name={axe | exception | error | jax, axis-name, collectives, api-misuse |
| preduced is a Invariant->Reduced collective. This means that | exception | error | jax, sharding, collectives, transpose, mesh |
| preduced input cannot be reduced across the axis_name provid | exception | error | jax, sharding, collectives, double-reduction, mesh |
| vary_unreduced_cast only accepts inputs that are varying. Go | exception | error | jax, named-axes, manual-parallelism, aval-validation |
| vary_unreduced_cast is a Varying->Unreduced collective. This | exception | error | jax, axis-name, named-axes |
| vary_unreduced_cast input cannot be unreduced across the axi | exception | error | jax, named-axes, duplicate-cast |
| reduced_vary_cast only accepts inputs that are reduced. Got | exception | error | jax, named-axes, manual-parallelism, aval-validation |
| reduced_vary_cast is a Reduced->Varying collective. This mea | exception | error | jax, axis-name, named-axes |
| reduced_vary_cast input cannot be varying across the axis_na | exception | error | jax, named-axes, manual-parallelism |
| {name} can only accept axis_name which corresponds to one of | exception | error | jax, pcast, named-axes, mixed-state |
| {axis_name=} must be a tuple or a str. Got {axis_name} | exception | error | jax, pcast, type-error, axis-name |
| Got unexpected `to` value. Allowed `to` values are: {_allowe | exception | error | jax, pcast, invalid-argument |
| Unsupported pcast from={from_}, {to=} | exception | error | jax, pcast, unsupported-transition, named-axes |
| Unsupported type for JSON serialization: {type(obj)} ({obj}) | exception | error | jax, json-serialization, async-collectives, config |
| async done op got {aval}, want core.AbstractFuture | exception | error | jax, async-collectives, future, type-error |
| {side} contracting dim {i} of size {size} must be divisible | exception | error | jax, scaled-dot, shape-validation, quantization |
| The ratio of {side} contracting dim {i} to its scale's dim s | exception | error | jax, scaled-dot, shape-validation |
| {side} dim {i} of size {size} does not match scale dim size | exception | error | jax, scaled-dot, shape-validation |
| All input tensors must have the same rank. Got lhs rank: {lh | exception | error | jax, scaled-dot, rank-validation |
| LHS and RHS must have the same number of batch dimensions, g | exception | error | jax, scaled-dot, dimension-numbers |
| dynamic_update_slice update shape must be smaller than opera | exception | error | jax, lax, dynamic-update-slice, shape-mismatch |
| start_indices arguments to dynamic_update_slice must be scal | exception | error | jax, lax, dynamic-update-slice, scalar-indices |
| index arguments to dynamic_update_slice must be integers of | exception | error | jax, lax, dtype, indices |
| indices must have an integer type | exception | error | jax, lax, gather, dtype |
| {name} in {op_name} op must be sorted; got {dims} | exception | error | jax, lax, gather, scatter, dimension-numbers |
| Invalid {name} set in {op_name} op; valid range is [0, {rank | exception | error | jax, lax, gather, scatter, dimension-out-of-range |
| Invalid {name} set in {op_name} op; valid range is [0, {rank | exception | error | jax, lax, gather, scatter, dimension-out-of-range |
| {name} in {op_name} op must not repeat; got: {dims}. | exception | error | jax, lax, gather, scatter, duplicate-dims |
| {name1} and {name2} in {op_name} op must be disjoint; got: { | exception | error | jax, lax, gather, scatter, dimension-numbers |
| Gather index leaf dimension must be within [0, rank(indices) | exception | error | jax, lax, gather, index-vector-dim |
| reduce_window expected init_values to be scalars but init_va | validation | error | jax, reduce-window, shape-validation, scalar |
| reduce_window batching is not implemented for initial values | exception | error | jax, vmap, batching, not-implemented, reduce-window |
| reduce_window jvp does not support non-zero init_value_tange | validation | error | jax, autodiff, jvp, reduce-window |
| Cannot lower effectful `reduce_window`. | exception | error | jax, effects, lowering, mlir, reduce-window |
| operand to reduce_window_sum must have a number dtype, got { | validation | error | jax, dtype, reduce-window, type-validation |
| reduce_window got the wrong number of window_dimensions for | validation | error | jax, shape-validation, reduce-window, windowing |
| reduce_window got inconsistent window_strides and window_dim | validation | error | jax, shape-validation, reduce-window, strides |
| reduce_window got inconsistent base_dilation and window_dime | validation | error | jax, shape-validation, reduce-window, dilation |
| reduce_window got inconsistent window_dilation and window_di | validation | error | jax, shape-validation, reduce-window, dilation |
| select_and_scatter got inconsistent window_strides and windo | validation | error | jax, shape-validation, select-and-scatter, pooling |
| select_and_gather_add tangents and operand shapes must match | validation | error | jax, autodiff, vjp, pooling, shape-validation |
| VJP not implemented for select_and_gather (MaxPool) with win | error_code | error | jax, autodiff, vjp, pooling, dilation, not-implemented |
| mulhi requires integer inputs, got {dtype} | validation | error | jax, dtype, integer, mulhi, type-validation |
| mulhi operands must have the same dtype, got {dtype} and {y_ | validation | error | jax, dtype, mulhi, type-mismatch |
| Dimension mismatch | validation | error | jax, numpy-reference, convolution, shape-validation |
| Wrong number of strides for spatial dimensions | validation | error | jax, numpy-reference, strides, shape-validation, convolution |
| Wrong number of pads for spatial dimensions | validation | error | jax, numpy-reference, padding, shape-validation, convolution |
| Expected mode of type `LayoutMode`. Got type: {type(mode)} | validation | error | jax, layout, type-validation, enum |
| Length of major_to_minor and the rank of the value should ma | validation | error | jax, layout, rank-mismatch, validation |
| Sharding has to be concrete when layout is of type {type(lay | validation | error | jax, layout, sharding, validation |
| Invalid value received for the layout argument. Expected val | validation | error | jax, layout, type-validation, isinstance-check |
| Invalid value received for the sharding argument. Expected v | validation | error | jax, sharding, type-validation |
| Cannot determine the ``__name__`` of the caller. | error_code | error | jax, lazy-loading, import-machinery, internal-api |
| module '{package_name}' has no attribute '{name}' | validation | error | jax, lazy-loading, attributeerror, module-structure |
| Attempting to set log level "{logging_level}" which isn't on | validation | error | jax, logging, config, valueerror |
| Please install the `etils[epath]` package to specify a cache | error_code | error | jax, compilation-cache, optional-dependency, cloud-storage |
| Please install the `filelock` package to set `jax_compilatio | error_code | error | jax, compilation-cache, optional-dependency, filelock |
| key cannot be empty | validation | error | jax, lru-cache, input-validation |
| Cannot update the mesh of the current resource environment. | validation | error | jax, mesh, axis-names, context-manager |
| When passing host local inputs to pjit, devices connected to | validation | critical | jax, mesh, multi-host, tpu, device-topology |
| axis_types passed to {name} must be of type `jax.sharding.Ax | validation | error | jax, mesh, axis-types, type-validation |
| Number of axis names should match the number of axis_types. | validation | error | jax, mesh, axis-types, shape-mismatch |
| Mesh axis names cannot be None. Got: {axis_names} | validation | error | jax, mesh, axis-names, input-validation |
| Mesh requires the ndim of its first argument (`devices`) to | validation | error | jax, mesh, shape-mismatch, input-validation |
| Mesh context manager is disabled. | error_code | error | jax, mesh, deprecation, context-manager, config-flag |
| AbstractMesh does not implement {name} | validation | error | jax, mesh, abstract-class, not-implemented |
| Expected mesh of type `jax.sharding.AbstractMesh`. Got type: | validation | error | jax, mesh, type-validation, abstract-mesh |
| use_abstract_mesh cannot change the size of the mesh. Got ne | validation | error | jax, mesh, abstract-mesh, size-mismatch |
| Failed to find assignment for logical_axis_index {logical_ax | exception | error | jax, mesh, tpu, topology, not-implemented |
| Reduced axes can only refer to mesh axes that is of type `Ex | validation | error | jax, sharding, mesh, distributed, named-sharding |
| Unknown algorithm '{algorithm}'. Expected 'fast' or 'stable' | validation | error | jax, nn, standardize, enum-argument |
| Expected num_classes to match the size of axis {axis}, but { | validation | error | jax, nn, one-hot, pmap, distributed |
| The input part of spec in out_sharding should match the spec | validation | error | jax, nn, one-hot, sharding, named-sharding |
| {name} ndim should be {len(shape)}, but got {t.ndim} | validation | error | jax, nn, attention, shape-validation |
| {name} dtype should be {dtype}, but got {t.dtype} | validation | error | jax, nn, attention, dtype-validation |
| {name} shape should be {shape}: but got {t.shape} | validation | error | jax, nn, attention, shape-validation |
| The number of query heads must be a multiple of key/value he | validation | error | jax, nn, attention, gqa, shape-validation |
| cuDNN doesn't support right window: {r_window} when causal m | validation | error | jax, nn, attention, cudnn, sliding-window, gpu |
| Unsupported implementation option: {implementation} | validation | error | jax, nn, attention, enum-argument |
| scaled_matmul requires all inputs to be 3-dimensional arrays | validation | error | jax, nn, matmul, float8, mxfp8, shape-validation |
| scaled_matmul requires inputs a and b to have matching batch | validation | error | jax, nn, matmul, float8, shape-validation |
| scaled_matmul requires scales to have matching batch (B) and | validation | error | jax, nn, matmul, float8, scales, shape-validation |
| scaled_matmul requires scales to match non-contract dimensio | validation | error | jax, nn, matmul, float8, scales, shape-validation |
| Unsupported mode: {mode} | validation | error | jax, nn, float8, config, enum-argument |
| Can't compute input and output sizes of a {len(shape)}-dimen | validation | error | jax, nn, initializers, shape-validation |
| invalid mode for variance scaling initializer: {mode} | exception | error | jax, nn, initializers, enum-argument |
| invalid distribution for variance scaling initializer: {dist | exception | error | jax, nn, initializers, enum-argument |
| orthogonal initializer requires at least a 2D shape | exception | error | jax, nn, initializers, shape-validation |
| Delta orthogonal initializer requires a 3D, 4D or 5D shape. | exception | error | jax, initializer, shape-validation |
| `fan_in` must be less or equal than `fan_out`. | exception | error | jax, initializer, shape-validation, fan-in-out |
| {api_version=!r} is not available; available versions are: { | exception | error | jax, array-api, version-mismatch |
| Currently, string arrays can only be made from NumPy arrays. | exception | error | jax, string-dtype, type-validation |
| Cannot make an array with dtype {dtype} from an object with | exception | error | jax, string-dtype, dtype-mismatch |
| ndmin {ndmin} cannot be greater than object's ndims {object. | exception | error | jax, string-dtype, shape-validation |
| array() takes at most 5 positional arguments but {len(args) | exception | error | jax, argument-validation, numpy-compat |
| array() got multiple values for argument '{name}' | exception | error | jax, argument-validation, numpy-compat, deprecation |
| Only implemented for order='K' | exception | error | jax, not-implemented, numpy-compat, memory-order |
| None is not a valid value for jnp.array | exception | error | jax, null-handling, data-validation |
| Unexpected input type for array: {type(object)} | exception | error | jax, type-validation, conversion |
| `{device_or_sharding = }` was passed to`canonicalize_or_get_ | exception | error | jax, device-validation, type-validation |
| jnp.asarray: cannot convert object of type {type(a)} to JAX | exception | error | jax, copy-semantics, device-placement, array-api |
| expected sequence object with len >= 0 or a single integer | exception | error | jax, shape-validation, generator |
| Cannot interpret '{dtype}' as a data type.\n\nDid you accide | exception | error | jax, shape-validation, dtype, typo-detection |
| Cannot interpret '{dtype}' as a data type.\n\nDid you accide | exception | error | jax, shape-validation, dtype, typo-detection |
| Cannot interpret '{dtype}' as a data type.\n\nDid you accide | exception | error | jax, shape-validation, dtype, typo-detection |
| Number of samples, {num}, must be non-negative. | exception | error | jax, linspace, argument-validation |
| Array.__contains__: unsupported operand type {type(other)}. | exception | error | jax, membership, typeerror, contains |
| Array.__contains__: search array must be one-dimensional, go | exception | error | jax, membership, shape-mismatch, contains |
| Array.__contains__: query value must be a scalar, got {query | exception | error | jax, membership, scalar-required, contains |
| No Python scalar type for {arr.dtype=} | exception | error | jax, dtype, item, extended-dtype |
| np.reshape order=A is not implemented. | exception | error | jax, reshape, not-implemented, numpy-compat |
| Unexpected value for 'order' argument: {order}. | exception | error | jax, reshape, invalid-argument, validation |
| stream argument of array.to_device() | exception | error | jax, device-transfer, stream, not-implemented |
| can only specify one unknown axis size with a `-1` value, go | exception | error | jax, reshape, invalid-shape |
| cannot reshape array of shape {arr.shape} (size {arr.size}) | exception | error | jax, reshape, size-mismatch |
| cannot reshape array of shape {arr.shape} (size {arr.size}) | exception | error | jax, reshape, size-mismatch |
| `type` argument of array.view() is not supported. | exception | error | jax, view, not-implemented, numpy-compat |
| When changing to a larger dtype, its size must be a divisor | exception | error | jax, view, dtype, alignment |
| JAX Arrays do not implement the arr.flat property: consider | exception | error | jax, flat, not-implemented, numpy-compat |
| dtypes of sorted_arr and query must match; got {sorted_arr_a | exception | error | jax, dtype, searchsorted, validation |
| invalid argument side={side!r}, expected 'left' or 'right' | exception | error | jax, searchsorted, argument-validation |
| invalid argument {method=}, expected one of {list(self.valid | exception | error | jax, searchsorted, argument-validation |
| batch dimension sizes must match; got {sorted_arr_aval.shape | exception | error | jax, shape-mismatch, searchsorted, batching |
| out_dtype should be an integer type; got {out_dtype} | exception | error | jax, dtype, searchsorted, argument-validation |
| size must be a positive integer; got {size=} | exception | error | jax, nonzero, argument-validation |
| out_dtype must be integer typed; got {out_dtype=} | exception | error | jax, dtype, nonzero, argument-validation |
| axes out of range for array with {a_aval.ndim} dimensions: | exception | error | jax, nonzero, axes, argument-validation |
| duplicate axes are not allowed: {axes=} | exception | error | jax, nonzero, axes, argument-validation |
| Expected {len(axes)} fill values, got {len(fill_value_avals) | exception | error | jax, nonzero, fill-value, argument-validation |
| Expected fill values to have dtype {out_dtype}, got {fill_va | exception | error | jax, nonzero, fill-value, dtype |
| fill_value shape {fv_aval.shape} is not broadcast-compatible | exception | error | jax, nonzero, fill-value, broadcasting |
| fill_value shape {fv_aval.shape} cannot be broadcast to batc | exception | error | jax, nonzero, fill-value, broadcasting |
| Run out of characters for batch dimension in einsum. | exception | error | jax, einsum, vmap, resource-limit |
| Transpose of Einsum with multiple linear inputs is not suppo | exception | error | jax, einsum, autodiff, not-implemented |
| Unsupported method: {method} | exception | error | jax, searchsorted, internal-api, argument-validation |
| fill_value tuple must have length equal to number of axes ({ | validation | error | jax, nonzero, fill-value, argument-validation |
| fill_value must be a scalar or tuple of scalars; got {fill_v | validation | error | jax, nonzero, fill-value, argument-validation |
| could not understand directive {directive!r} | validation | error | jax, r-c-index-tricks, directive, numpy-compat |
| string directive must be placed at the beginning | validation | error | jax, r-c-index-tricks, directive, numpy-compat |
| Indexer must have integer or boolean type, got indexer with | validation | error | jax, indexing, dtype |
| JAX does not support string indexing; got {idx=} | validation | error | jax, indexing, string-index |
| Indexer must have integer or boolean type, got indexer with | validation | error | jax, indexing, dtype, sequence |
| Indexer must have integer or boolean type, got indexer with | validation | error | jax, indexing, scalar, dtype |
| only integers, slices (`:`), ellipsis (`...`), newaxis (`Non | validation | error | jax, indexing, type-error |
| Unrecognized index type: {typ} | validation | error | jax, internal, invariant |
| an index can only have a single ellipsis ('...') | validation | error | jax, indexing, ellipsis |
| Too many indices: array is {len(shape)}-dimensional, but {to | validation | error | jax, indexing, rank, shape-mismatch |
| index {i} out of bounds for axis {axis} with size {size} ({n | validation | error | jax, indexing, out-of-bounds |
| Array slice indices must have static start/stop/step to be u | validation | error | jax, jit, tracing, dynamic-slice |
| boolean index did not match shape of indexed array in index | validation | error | jax, boolean-mask, shape-mismatch, indexing |
| mode='slice' is not valid for polymorphic shapes. | validation | error | jax, polymorphic-shapes, export, slicing |
| static_slice requires mode='promise_in_bounds' or mode='clip | validation | error | jax, indexing, mode, api-misuse |
| static_slice with partial slices does not support nontrivial | validation | error | jax, sharding, multi-device, slicing |
| static_slice: indices must be static scalars or slices. Got | validation | error | jax, static-slice, advanced-indexing |
| static_slice: unrecognized index {pidx.index} at position {p | validation | error | jax, internal, invariant |
| index is out of bounds for axis {axis} with size 0 | validation | error | jax, empty-array, out-of-bounds, indexing |
| static_slice: unrecognized index {pidx.index} | validation | error | jax, internal, invariant |
| dynamic_slice requires mode='promise_in_bounds' or mode='cli | validation | error | jax, dynamic-slice, mode, api-misuse |
| dynamic_slice with partial slices does not support nontrivia | validation | error | jax, sharding, dynamic-slice, multi-device |
| dynamic_slice: only unit steps supported in slice. Got {pidx | validation | error | jax, indexing, slice, stride, dynamic-slice |
| dynamic_slice: only scalar indices allowed. Got index of typ | validation | error | jax, indexing, scalar-index, dynamic-slice |
| dynamic_slice: indices must be scalars or slices. Got index | validation | error | jax, indexing, boolean-index, dynamic-slice |
| dynamic_slice: unrecognized index {pidx.index} at position { | validation | error | jax, indexing, unsupported-index, dynamic-slice |
| dynamic_slice: unrecognized index {pidx.index} | validation | error | jax, indexing, internal, dynamic-slice |
| The 'out' argument to jnp.take is not supported. | validation | error | jax, numpy-compat, out-arg, immutable |
| The 'raise' mode to jnp.take is not supported. | validation | error | jax, take, mode, bounds-check |
| Invalid mode '{mode}' for np.take | validation | error | jax, take, invalid-argument, mode |
| Cannot do a non-empty jnp.take() from an empty axis. | validation | error | jax, take, empty-array, out-of-bounds |
| take_along_axis indices must be of integer type, got {index_ | validation | error | jax, take-along-axis, dtype, integer-index |
| take_along_axis indices must be 1D if axis=None, got shape { | validation | error | jax, take-along-axis, shape-mismatch, axis |
| indices and arr must have the same number of dimensions; {} | validation | error | jax, take-along-axis, ndim, shape-mismatch |
| One Hot indexing is only supported for up to 50 leading dime | exception | error | jax, take-along-axis, rank-limit, einsum |
| jax.numpy.put_along_axis cannot modify arrays in-place, beca | exception | error | jax, put-along-axis, immutability, inplace |
| put_along_axis arguments 'arr' and 'indices' must have same | exception | error | jax, put-along-axis, ndim, shape-mismatch |
| put_along_axis argument 'values' must be broadcastable to 'i | exception | error | jax, put-along-axis, broadcasting, shape-mismatch |
| Expected strategy to be IndexingStrategy; got {strategy} | exception | error | jax, internal, indexing-strategy, type-error |
| fill_value argument to indexed get() must be a scalar | exception | error | jax, indexing, fill-value, scalar |
| the only valid string value of `left` is 'extrapolate', but | exception | error | jax, interp, invalid-argument, sentinel-value |
| the only valid string value of `right` is 'extrapolate', but | exception | error | jax, interp, invalid-argument, sentinel-value |
| jnp.interp: complex x values not supported. | exception | error | jax, interp, complex-dtype, unsupported-operation |
| period must be a scalar; got {period} | exception | error | jax, interp, period, shape-validation |
| size and fill_value arguments cannot be used in three-term w | exception | error | jax, where, invalid-argument, api-misuse |
| Either both or neither of the x and y arguments should be pr | exception | error | jax, where, missing-argument |
| condlist must have length equal to choicelist ({} vs {}) | exception | error | jax, select, length-mismatch |
| condlist must be non-empty | exception | error | jax, select, empty-list |
| x argument to bincount must have an integer type; got {x.dty | exception | error | jax, bincount, dtype, typeerror |
| only 1-dimensional input supported. | exception | error | jax, bincount, ndim, shape-validation |
| type of weights must match type of x. Got typeof(x)={core.ty | exception | error | jax, bincount, weights, sharding, shape-mismatch |
| jax.numpy.{op} with a symbolic number of sections is not sup | exception | error | jax, split, symbolic-dim, tracing |
| array split does not result in an equal division: rest is {r | exception | error | jax, split, uneven-division |
| No input was provided to the clip function. | exception | error | jax, clip, missing-argument |
| Clip received a complex value either through the input or th | exception | error | jax, clip, complex-dtype, unsupported-operation |
| The 'out' argument to jnp.round is not supported. | exception | error | jax, round, out-parameter, notimplementederror |
| integer np.round not implemented for decimals < 0 | exception | error | jax, round, integer-dtype, notimplementederror |
| fill_value must be a scalar or a tuple of length {arr.ndim}; | exception | error | jax, nonzero, fill-value, scalar-validation |
| jnp.unwrap does not support complex inputs. | exception | error | jax, unwrap, complex-dtype, unsupported-operation |
| `{name}` entries must be the same shape: {nvals} | exception | error | jnp-pad, shape-mismatch, numpy |
| jnp.pad: {name} with {nd=} has unsupported shape {nvals.shap | exception | error | jnp-pad, invalid-argument-shape |
| Cannot apply '{}' padding to empty axis | exception | error | jnp-pad, empty-axis, reflect, wrap |
| jnp.pad: constant_values has unsupported shape {constant_val | exception | error | jnp-pad, constant-values, invalid-argument-shape |
| Shape polymorphism is supported for jnp.pad with 'reflect' o | exception | error | jnp-pad, shape-polymorphism, symbolic-dimensions, reflect |
| stat_length of 0 yields no value for padding | exception | error | jnp-pad, stat-length, statistical-mode |
| Expected pad_width to have shape {(nd, 2)}; got {pad_width_a | exception | error | jnp-pad, pad-width, shape-validation |
| index can't contain negative values | exception | error | jnp-pad, negative-pad-width |
| `pad_width` must be of integral type. | exception | error | jnp-pad, type-error, integral-type |
| Unimplemented padding mode '{}' for np.pad. | exception | error | jnp-pad, unsupported-mode, not-implemented |
| unsupported keyword arguments for mode '{}': {} | exception | error | jnp-pad, unsupported-kwarg, keyword-arguments |
| Need at least one array to stack. | validation | error | jnp-stack, empty-sequence, value-error |
| The 'out' argument to jnp.stack is not supported. | exception | error | jnp-stack, out-argument, immutable-arrays, numpy-parity |
| Need at least one array to concatenate. | validation | error | jnp-concatenate, empty-sequence, value-error |
| Zero-dimensional arrays cannot be concatenated. | validation | error | jnp-concatenate, wrong-argument-form |
| The 'out' argument to jnp.choose is not supported. | exception | error | jnp-choose, out-argument, immutable-arrays, numpy-parity |
| `a` array must be integer typed | validation | error | jax, numpy, dtype, valueerror, choose |
| invalid entry in choice array | validation | error | jax, choose, index-out-of-range, valueerror |
| mode={mode!r} not understood. Must be 'raise', 'wrap', or 'c | validation | error | jax, choose, invalid-argument, valueerror |
| jax.numpy.block does not allow tuples, got {} | validation | error | jax, block, tuple-vs-list, valueerror |
| jax.numpy.block does not allow empty list arguments | validation | error | jax, block, empty-list, valueerror |
| Mismatched list depths in jax.numpy.block | validation | error | jax, block, shape-mismatch, valueerror |
| jnp.fromfile() is not implemented because it may be non-pure | exception | error | jax, fromfile, notimplemented, io, jit-purity |
| jnp.fromiter() is not implemented because it may be non-pure | exception | error | jax, fromiter, notimplemented, io, jit-purity |
| k must be a scalar integer; got {k} | validation | error | jax, eye, dtype, scalar, valueerror |
| negative dimensions are not allowed, got {N} and {M} | validation | error | jax, eye, negative-dimension, valueerror |
| jax.numpy.arange: arguments must be scalars; got {name}={val | validation | error | jax, arange, scalar-requirement, valueerror |
| Passing complex start/stop/step to jnp.arange is no longer s | validation | error | jax, arange, complex, breaking-change, v0-10 |
| In arange with non-constant arguments all of start, stop, an | validation | error | jax, arange, symbolic-dim, shape-polymorphism, valueerror |
| arange has step == 0 | validation | error | jax, arange, zero-step, valueerror |
| jax.numpy.meshgrid only supports copy=True | validation | error | jax, meshgrid, copy-semantics, valueerror |
| Valid values for indexing are 'xy' and 'ij', got {indexing} | validation | error | jax, meshgrid, indexing, invalid-argument |
| Unsupported input type to jax.numpy.i0: {x_arr.dtype} | validation | error | jax, i0, bessel, dtype, complex |
| np.delete(arr, obj): for boolean indices, obj must be one-di | exception | error | jax, numpy-compat, boolean-mask, shape-mismatch |
| np.delete(arr, obj): got obj.dtype={obj_array.dtype}; must b | exception | error | jax, dtype, indexing |
| jnp.insert(): obj must be a slice, a one-dimensional array, | exception | error | jax, numpy-compat, insert, ndim |
| jnp.insert(): index array must be integer typed; got {obj} | exception | error | jax, insert, dtype, indexing |
| function is not returning an array of the correct shape | exception | error | jax, apply-along-axis, shape-mismatch |
| Dimension must be either 2 or 3 for cross product | exception | error | jax, cross-product, shape-validation |
| x must be a one-dimensional array | exception | error | jax, vander, shape-validation |
| N must be nonnegative | exception | error | jax, vander, argument-validation |
| The 'out' argument to jnp.argmax is not supported. | exception | error | jax, argmax, unsupported-argument, numpy-compat |
| attempt to get argmax of an empty sequence | exception | error | jax, argmax, empty-array |
| The 'out' argument to jnp.argmin is not supported. | exception | error | jax, argmin, unsupported-argument, numpy-compat |
| attempt to get argmin of an empty sequence | exception | error | jax, argmin, empty-array |
| The 'out' argument to jnp.nanargmax is not supported. | exception | error | jax, nanargmax, unsupported-argument, numpy-compat |
| The 'out' argument to jnp.nanargmin is not supported. | exception | error | jax, nanargmin, unsupported-argument, numpy-compat |
| 'shift' and 'axis' arguments to roll must be scalars or 1D a | exception | error | jax, roll, shape-validation |
| {start=} must satisfy {-a_ndim}<=start<={a_ndim} | exception | error | jax, rollaxis, argument-validation |
| Expected an input array of integer or boolean data type | exception | error | jax, packbits, dtype-validation |
| 'order' must be either 'little' or 'big' | exception | error | jax, packbits, bitorder, argument-validation |
| Expected an input array of unsigned byte data type | exception | error | jax, unpackbits, dtype-validation |
| Arguments to jax.numpy.gcd must be integers. | exception | error | jax, numpy, dtype-validation, integer-required |
| Arguments to jax.numpy.lcm must be integers. | exception | error | jax, numpy, dtype-validation, integer-required |
| The 'out' argument to jnp.compress is not supported. | exception | error | jax, unsupported-argument, numpy-compat, out-parameter |
| condition contains entries that are out of bounds | exception | error | jax, out-of-bounds, compress, length-mismatch |
| size must be positive and not greater than the size of the a | exception | error | jax, size-validation, out-of-range |
| y has more than 2 dimensions | exception | error | jax, statistics, covariance, shape-validation |
| m has more than 2 dimensions | exception | error | jax, statistics, covariance, ndim-validation |
| cov: dtype must be a subclass of float or complex; got {dtyp | exception | error | jax, covariance, dtype-validation |
| cannot handle multidimensional fweights | exception | error | jax, covariance, weights, shape-validation |
| incompatible numbers of samples and fweights | exception | error | jax, covariance, weights, length-mismatch |
| fweights must be integer. | exception | error | jax, covariance, weights, dtype-validation |
| cannot handle multidimensional aweights | exception | error | jax, covariance, weights, shape-validation |
| incompatible numbers of samples and aweights | exception | error | jax, covariance, weights, length-mismatch |
| corrcoef: dtype must be a subclass of float or complex; got | exception | error | jax, corrcoef, dtype-validation |
| with {nc} condition(s), either {nc} or {nc+1} functions are | exception | error | jax, piecewise, list-length-mismatch |
| {}-dimensional array given. Array must be at least two-dimen | exception | error | jax, linalg, matrix-power, ndim-validation |
| Last 2 dimensions of the array must be square | exception | error | jax, linalg, matrix-power, square-matrix-required |
| matrix_transpose requires at least 2 dimensions; got {ndim=} | exception | error | jax, numpy, shape-validation, linalg |
| Invalid order '{ord}' for vector norm. | exception | error | jax, linalg, norm, type-validation |
| ind must be a positive integer; got {ind=} | exception | error | jax, scipy, linalg, tensor, argument-validation |
| tensorinv is only possible when the product of the first `in | exception | error | jax, scipy, linalg, tensor, shape-validation |
| After moving axes to end, leading shape of a must match shap | exception | error | jax, numpy, linalg, tensor, shape-validation |
| Input arrays must have prod(a.shape[:b.ndim]) == prod(a.shap | exception | error | jax, numpy, linalg, tensor, shape-validation |
| multi_dot requires at least two arrays; got len(arrays)={len | exception | error | jax, numpy, linalg, matmul, argument-validation |
| multi_dot: input arrays must all be two-dimensional, except | exception | error | jax, numpy, linalg, matmul, shape-validation |
| multi_dot: last dimension of each array must match first dim | exception | error | jax, numpy, linalg, matmul, shape-mismatch |
| jnp.linalg.cond: input array must be at least 2D; got {arr.s | exception | error | jax, numpy, linalg, condition-number, shape-validation |
| jnp.linalg.cond: input array must not be empty; got {arr.sha | exception | error | jax, linalg, condition-number, empty-array, edge-case |
| jnp.linalg.cond: for {p=}, array must be square; got {arr.sh | exception | error | jax, numpy, linalg, condition-number, shape-validation |
| Input must be a rank-1 array. | validation | error | jax, numpy, polynomial, roots, shape-validation |
| expected deg >= 0 | validation | error | jax, numpy, polynomial, polyfit, argument-validation |
| expected 1D vector for x | validation | error | jax, numpy, polynomial, polyfit, shape-validation |
| expected non-empty vector for x | validation | error | jax, numpy, polynomial, polyfit, empty-array, edge-case |
| expected 1D or 2D array for y | validation | error | jax, numpy, polynomial, polyfit, shape-validation |
| expected x and y to have same length | validation | error | jax, numpy, polynomial, polyfit, shape-mismatch |
| expected a 1-d array for weights | validation | error | jax, numpy, polynomial, polyfit, weights, shape-validation |
| expected w and y to have the same length | validation | error | jax, numpy, polynomial, polyfit, weights, length-mismatch |
| the number of data points must exceed order to scale the cov | validation | error | jax, polyfit, covariance, insufficient-data |
| input must be 1d or non-empty square 2d array. | validation | error | jax, numpy-poly, input-shape |
| Order of integral must be positive (see polyder) | validation | error | jax, polyint, argument-validation |
| k must be a scalar or a rank-1 array of length 1 or m. | validation | error | jax, polyint, shape-mismatch |
| Order of derivative must be positive | validation | error | jax, polyder, argument-validation |
| jnp.{name}: where must be None or a boolean array; got {wher | validation | error | jax, reductions, where-mask, dtype |
| The 'out' argument to jnp.{name} is not supported. | validation | error | jax, reductions, out-parameter, immutable-arrays |
| reduction operation {name} does not have an identity, so to | validation | error | jax, reductions, where-mask, identity |
| zero-size array to reduction operation {name} which has no i | validation | error | jax, reductions, empty-array |
| Named reductions not implemented for jnp.{name}() | validation | error | jax, named-axes, pmap, reductions |
| initial value must be a scalar. Got array of shape {initial_ | validation | error | jax, reductions, initial-value, shape-mismatch |
| duplicate value in 'axis': {axis} | validation | error | jax, reductions, axis, duplicate |
| integer argument required; got dtype={arr.dtype} | validation | error | jax, reductions, dtype, internal |
| The 'out' argument to jnp.logaddexp.reduce is not supported. | validation | error | jax, logsumexp, out-parameter |
| The 'out' argument to jnp.logaddexp2.reduce is not supported | validation | error | jax, logaddexp2, out-parameter |
| The 'out' argument to jnp.mean is not supported. | validation | error | jax, mean, out-parameter |
| Axis must be specified when shapes of a and weights differ. | validation | error | jax, average, weights, shape-mismatch |
| Shape of weights must be consistent with shape of a along sp | validation | error | jax, average, weights, shape-mismatch |
| ddof and correction can't be provided simultaneously. | validation | error | jax, var, ddof, parameter-conflict |
| The 'out' argument to jnp.var is not supported. | validation | error | jax, var, out-parameter |
| jax.numpy.var does not yet support real dtype parameters whe | validation | error | jax, numpy, variance, complex-numbers, dtype |
| dtype argument to jnp.std must be inexact; got {dtype} | validation | error | jax, numpy, std, dtype-validation |
| The 'out' argument to jnp.std is not supported. | validation | error | jax, numpy, out-argument, immutable-arrays |
| The 'out' argument to jnp.ptp is not supported. | validation | error | jax, numpy, ptp, out-argument |
| The 'out' argument to jnp.nanmean is not supported. | validation | error | jax, numpy, nanmean, out-argument |
| The 'out' argument to jnp.nanvar is not supported. | validation | error | jax, numpy, nanvar, out-argument |
| The 'out' argument to jnp.nanstd is not supported. | validation | error | jax, numpy, nanstd, out-argument |
| The 'out' argument to jnp.{name} is not supported | validation | error | jax, numpy, cumsum, cumprod, out-argument |
| The input must be non-scalar to take a cumulative sum, howev | validation | error | jax, numpy, cumulative-sum, scalar-input |
| The input array has rank {x.ndim}, however axis was not set | validation | error | jax, numpy, cumulative-sum, axis-argument |
| The input must be non-scalar to take a cumulative product, h | validation | error | jax, numpy, cumulative-prod, scalar-input |
| jax.numpy.quantile does not support overwrite_input=True or | validation | error | jax, numpy, quantile, out-argument, overwrite-input |
| jax.numpy.nanquantile does not support overwrite_input=True | validation | error | jax, numpy, nanquantile, out-argument, overwrite-input |
| method can only be 'linear', 'lower', 'higher', 'midpoint', | validation | error | jax, numpy, quantile, invalid-argument |
| Weights cannot be complex types. | validation | error | jax, numpy, quantile, weights, complex-numbers |
| {method} doesn't support weights. Only method 'inverted_cdf' | validation | error | jax, numpy, quantile, weights, not-implemented |
| Weights shape must match 'a' shape when axis is None. | validation | error | jax, numpy, quantile, weights, shape-mismatch |
| Weights shape {weights.shape} must match reduction axes {tup | validation | error | jax, numpy, quantile, weights, shape-mismatch |
| 'order' argument to argsort is not supported. | validation | error | jax, argsort, numpy-compat, unsupported-argument, sorting |
| jnp.partition for complex dtype is not implemented. | validation | error | jax, partition, complex-dtype, not-implemented, sorting |
| jnp.argpartition for complex dtype is not implemented. | validation | error | jax, argpartition, complex-dtype, not-implemented, sorting |
| need sequence of keys with len > 0 in lexsort | validation | error | jax, lexsort, empty-input, validation, sorting |
| all keys need to be the same shape | validation | error | jax, lexsort, shape-mismatch, validation, sorting |
| top_k is not compatible with complex inputs. | validation | error | jax, top-k, complex-dtype, value-error, sorting |
| mode must be 'largest' or 'smallest', got {mode!r} | validation | error | jax, top-k, invalid-argument, value-error, enum-validation |
| Number of tensordot axes (axes {}) exceeds input ranks ({} a | validation | error | jax, tensordot, shape-mismatch, numpy |
| tensordot requires axes lists to have equal length, got {} a | validation | error | jax, tensordot, axes-mismatch, numpy |
| tensordot requires both axes lists to be either ints, tuples | validation | error | jax, tensordot, type-error, axes |
| tensordot axes argument must be an int, a pair of ints, or a | validation | error | jax, tensordot, invalid-argument, numpy |
| The 'out' argument to jnp.outer is not supported. | validation | error | jax, outer, immutable-arrays, numpy-compat |
| out argument of {self} | validation | error | jax, ufunc, immutable-arrays, numpy-compat |
| where argument of {self} | validation | error | jax, ufunc, where-mask, numpy-compat |
| reduce only supported for binary ufuncs | validation | error | jax, ufunc, reduce, api-misuse |
| reduce only supported for functions returning a single value | validation | error | jax, ufunc, reduce, api-misuse |
| out argument of {self.__name__}.reduce() | validation | error | jax, ufunc, reduce, immutable-arrays |
| reduction operation {self.__name__!r} does not have an ident | validation | error | jax, ufunc, reduce, where-mask |
| where argument must have dtype=bool; got dtype={lax._dtype(w | validation | error | jax, ufunc, reduce, dtype, where-mask |
| tuple of axes | exception | error | jax, ufunc, reduce, multi-axis |
| zero-size array to reduction operation {self.__name__} which | exception | error | jax, ufunc, reduce, empty-array |
| accumulate only supported for binary ufuncs | exception | error | jax, ufunc, accumulate, api-misuse |
| accumulate only supported for functions returning a single v | exception | error | jax, ufunc, accumulate, api-misuse |
| out argument of {self.__name__}.accumulate() | exception | error | jax, ufunc, accumulate, immutable-arrays |
| accumulate does not allow multiple axes | exception | error | jax, ufunc, accumulate, axis |
| Because JAX arrays are immutable, jnp.ufunc.at() cannot oper | exception | error | jax, ufunc, scatter-add, immutable-arrays |
| reduceat only supported for binary ufuncs | exception | error | jax, ufunc, reduceat, api-misuse |
| num_segments must be non-negative. | validation | error | jax, segment-reduction, argument-validation, valueerror |
| unreduced for prod is not yet supported. | exception | error | jax, pallas-sharding, notimplementederror, distributed |
| unreduced for max is not yet supported. | exception | error | jax, sharding, notimplementederror, distributed |
| unreduced for min is not yet supported. | exception | error | jax, sharding, notimplementederror, distributed |
| Compiler params for platform {platform} cannot be used for { | validation | error | jax, pallas, backend-mismatch, gpu, tpu |
| MemoryRef type must be a ShapedArray, got {type(self.inner_a | validation | error | jax, pallas, aval, type-validation |
| Memory space {self.memory_space} is not supported by mesh {s | validation | error | jax, pallas, memory-space, mesh, gpu, tpu |
| Axis {axis} is out of bounds for grid {self.grid} | validation | error | jax, pallas, grid, index-out-of-bounds |
| Unsupported block dimension type: {type(dim)}. Allowed types | validation | error | jax, pallas, block-spec, type-validation |
| Unsupported block shape type: {type(dim)} | validation | error | jax, pallas, block-spec, internal |
| Array type must have a `shape` attribute, but got {type(arra | validation | error | jax, pallas, block-spec, aval |
| Block shape for {origin} (= {block_shape}) must have the sam | validation | error | jax, pallas, block-spec, shape-mismatch |
| shape polymorphism for Pallas does not support dynamically-s | validation | error | jax, pallas, shape-polymorphism, export, experimental |
| Index map function {debug_info.func_src_info} for {origin} m | validation | error | jax, pallas, block-spec, index-map |
| index_map returned a value of type {type(idx_aval)} at posit | validation | error | jax, pallas, index-map, block-spec |
| index_map returned a value of type {type(idx_aval)} at posit | validation | error | jax, pallas, index-map, scalar |
| Index map function {debug_info.func_src_info} for {origin} m | validation | error | jax, pallas, index-map, dtype |
| Index map function {debug_info.func_src_info} for {origin} m | validation | error | jax, pallas, index-map, closure |
| BlockMapping for {self.origin} has captured constants: {self | validation | error | jax, pallas, block-mapping, internal, closure |
| Cannot eagerly run with_memory_space_constraint. | validation | error | jax, pallas, eager-execution, primitive, memory-space |
| with_memory_space_constraint only supports arrays. | validation | error | jax, pallas, abstract-value, shapedarray, type-mismatch |
| When `check_vma=True` on `jax.shard_map`, `manual_axis_type` | validation | error | jax, shard-map, vma, shapedtypestruct, mesh, tpu, config-flag |
| Invalid out_shape type: {type(out_shape)} | validation | error | jax, pallas, out-shape, shapedtypestruct, input-validation |
| Unmatched parenthesis in {s!r} | validation | error | jax, einshape, einsum-notation, parse-error, parentheses |
| Equation must contain exactly one '->' | validation | error | jax, einshape, einsum-notation, equation-format, parse-error |
| Inconsistent size for {name}: {dim_sizes[name]} vs {shape_va | validation | error | jax, einshape, einsum-notation, dimension-mismatch, shape-validation |
| Size mismatch for group {group}: expected {shape_val}, got { | validation | error | jax, einshape, einsum-notation, shape-mismatch, grouped-reshape |
| get not supported yet | exception | error | jax, pallas, get, indexer, not-implemented |
| get not supported yet for block shape {b} | exception | error | jax, pallas, element-indexing, get, not-implemented |
| Concatenation with Element indexing is not yet supported. | exception | error | jax, pallas, concatenate, element-indexing, not-implemented |
| Shape along concat dimension {dimension} must be divisible b | validation | error | jax, pallas, concatenate, block-divisibility, valueerror |
| Stack with Element indexing is not yet supported. | exception | error | jax, pallas, stack, element-indexing, not-implemented |
| Stacking only supported when the block size along the stack | exception | error | jax, pallas, fusion, block-spec, shape-mismatch |
| Sum of sizes {n} must be equal to dimension {axis} of the op | exception | error | jax, pallas, lax-split, shape-mismatch, validation |
| Cannot permute last two dimensions with leading dimensions. | exception | error | jax, pallas, transpose, permutation, not-implemented |
| tile with non-int block dimensions not supported yet | exception | error | jax, pallas, tile, block-spec, not-implemented |
| Block size must be a multiple of the input size. Got block { | exception | error | jax, pallas, tile, shape-mismatch, divisibility |
| Every block dimension must be either a multiple or factor of | exception | error | jax, pallas, tile, divisibility, shape-mismatch |
| bitcast_convert_type with different bitwidths not supported | exception | error | jax, pallas, bitcast, dtype, not-implemented |
| Cannot pull iota along dimension {dimension} with None block | exception | error | jax, pallas, iota, block-spec, validation |
| reshape not supported yet: {aval_in}, {aval_out} | exception | error | |
| Expected {len(update_block_spec.block_shape)} start indices, | validation | error | |
| Parameter {key} is not a Jaxpr or sequence of Jaxprs: {value | validation | error | jax, pallas, jaxpr, control-flow, interpreter |
| run_scoped interpret rule does not support collective axes | exception | error | jax, pallas, collectives, interpreter, notimplemented |
| Padding with aliasing not supported. | exception | error | jax, pallas, aliasing, padding, interpreter |
| Acc ref must be at least 2D, got shape {shape} | validation | error | jax, pallas, tpu, mosaic, accumulator, shape |
| Acc ref dtype must be float32 or int32, got {dtype} | validation | error | jax, pallas, tpu, accumulator, dtype |
| Accumulators are not available on TPU {info.chip_version} | validation | error | jax, pallas, tpu, hardware, accumulator |
| mxu_id must be in [0, {info.num_mxus}), got {mxu_id=} | validation | error | jax, pallas, tpu, accumulator, index-out-of-range |
| The minor dimension size of an accumulator ref must be {info | validation | error | jax, pallas, tpu, accumulator, shape |
| The product of the major dimensions must be a multiple of {i | validation | error | jax, pallas, tpu, accumulator, shape, alignment |
| You can't use two different TensorCoreMeshes. | validation | error | jax, pallas, tpu, mesh, tensorcore, multi-core |
| cannot specify both devices and num_cores | validation | error | jax, pallas, tpu, mesh, mutually-exclusive-args |
| Unsupported core type: {core_type} | validation | error | jax, pallas, tpu, memory-space, lowering, version-skew |
| Invalid memory space: {memory_space!r} | validation | error | jax, pallas, tpu, memory-space, validation |
| More than one location found in string: {location_string} | validation | error | jax, pallas, tpu, error-handling, parsing |
| Could not find location in string {location_string} | validation | warning | jax, pallas, tpu, error-handling, parsing |
| Unknown action: {action} | validation | error | jax, pallas, tpu, async-copy, internal-api |
| Advanced indexers are not supported on TPU | validation | error | jax, pallas, tpu, interpreter, advanced-indexing |
| Revisited block {output_ranges[i]} of output {i} in iteratio | validation | critical | jax, pallas, tpu, interpreter, overlapping-writes, grid |
| Out-of-bounds read of ({device_id} {local_core_id} {memory_s | validation | error | jax, pallas, tpu, out-of-bounds, interpret-mode |
| Out-of-bounds block index {block_indices} for input "{input_ | validation | error | jax, pallas, tpu, blockspec, out-of-bounds |
| Out-of-bounds write of ({device_id} {local_core_id} {memory_ | validation | error | jax, pallas, tpu, out-of-bounds, write |
| Out-of-bounds block index {block_indices} for output "{outpu | validation | error | jax, pallas, tpu, blockspec, out-of-bounds |
| Out-of-bounds swap of ({device_id} {local_core_id} {memory_s | validation | error | jax, pallas, tpu, atomics, out-of-bounds |
| Out-of-bounds masked swap of ({device_id} {local_core_id} {m | validation | error | jax, pallas, tpu, atomics, mask, out-of-bounds |
| {primitive_name}: Buffers with a memory space of HBM or ANY | validation | error | jax, pallas, tpu, hbm, vmem, dma |
| masked load_p | validation | error | jax, pallas, tpu, interpret-mode, not-implemented, masked-load |
| run_scoped_p with collective axes is not supported | validation | error | jax, pallas, tpu, interpret-mode, not-implemented, collectives |
| Non-decrementing wait is not supported. | validation | error | jax, pallas, tpu, semaphore, interpret-mode, not-implemented |
| Unsupported block dim type: {type(b)} | validation | error | jax, pallas, blockspec, type-error |
| Dimensions with parallel semantics must form a prefix of the | validation | error | jax, pallas, tpu, grid, dimension-semantics |
| Unsupported memory space: {x.memory_space} | validation | error | jax, pallas, memory-space, not-implemented |
| Aliasing of scalar prefetch arguments is not currently suppo | validation | error | jax, pallas, tpu, aliasing, scalar-prefetch, interpret-mode |
| Kernel input {j} in HBM but does not have trivial BlockSpec. | validation | error | jax, pallas, tpu, hbm, blockspec, interpret-mode |
| Kernel output {j} in HBM but does not have trivial BlockSpec | validation | error | jax, pallas, tpu, hbm, blockspec, interpret-mode |
| Number of cores or threads must be at least 1, but got {self | validation | error | jax, pallas, tpu, config-validation, interpret-mode |
| Vector clock size must be at least 1, but got {self.vector_c | validation | error | jax, pallas, tpu, config-validation, race-detection |
| Vector clock size ({self.vector_clock_size}) must be greater | validation | error | jax, pallas, mosaic, tpu, interpret-mode, vector-clock, config |
| Logical shape {self.logical_shape} cannot be bigger than con | validation | error | jax, pallas, mosaic, shared-memory, shape-mismatch, interpret-mode |
| Range {rnge} is entirely out of bounds for shape {self.shape | validation | error | jax, pallas, mosaic, index-error, out-of-bounds, interpret-mode |
| Range {rnge} is (at least partially) out of bounds for alloc | exception | error | jax, pallas, mosaic, index-error, out-of-bounds, write |
| Attempting to deallocate allocation with key `{key}` that is | exception | error | jax, pallas, mosaic, shared-memory, key-collision, interpret-mode |
| Attempting to get contents of allocation with key `{key}` th | exception | error | jax, pallas, mosaic, shared-memory, key-collision, interpret-mode |
| Attempting to store into allocation with key `{key}` that is | exception | error | jax, pallas, mosaic, shared-memory, key-collision, interpret-mode |
| Attempting to swap into allocation with `key` {key} that is | exception | error | jax, pallas, mosaic, shared-memory, key-collision, swap |
| Semaphore {sem_id} occurs as both fixed-id and internal. | exception | error | jax, pallas, mosaic, semaphore, id-collision, interpret-mode |
| Computation failed on {self._failed_thread} with exception:\ | exception | error | jax, pallas, mosaic, interpret-mode, threading, error-propagation |
| {uninitialized_memory} + {str(dtype)} | exception | error | jax, pallas, mosaic, dtype, not-implemented, interpret-mode |
| {axis} mixes JAX mesh and Pallas mesh grid axes | exception | error | jax, pallas, mesh, device-id, interpret-mode, not-implemented |
| {non_mesh_axes} | exception | error | jax, pallas, mesh, device-id, axis-name, interpret-mode |
| Unsupported device ID type: {device_id_type} | exception | error | jax, pallas, device-id, version-mismatch, interpret-mode |
| Unsupported range type: {type(r)}. | exception | error | jax, pallas, mosaic, indexing, unsupported-type |
| position {position} is out of range for clock {self.clock} | exception | error | jax, pallas, mosaic, vector-clock, race-detection, config |
| Too many dynamic shapes in the input. Mosaic currently only | validation | error | jax, pallas, mosaic, dynamic-shapes, limit-exceeded, lowering |
| program id was requested but no grid was provided. | exception | error | jax, pallas, mosaic, grid, program-id, lowering |
| Invalid axis {axis} for num_programs | exception | error | jax, pallas, mosaic, grid, axis, lowering |
| The Pallas TPU lowering currently supports in memory space A | validation | error | jax, pallas, tpu, block-mapping, memory-space |
| The Pallas TPU lowering currently requires that the last two | validation | error | jax, pallas, tpu, alignment, dynamic-shapes |
| The Pallas TPU lowering currently requires that rank 1 block | validation | error | jax, pallas, tpu, block-shape, rank-1 |
| Pallas TPU requires a recent libtpu version (at least 0.0.46 | exception | critical | jax, pallas, tpu, libtpu, version-mismatch |
| Non-trivial windowing is not supported for grid-free pallas_ | exception | error | jax, pallas, tpu, grid, windowing |
| Unsupported block dimension type: {type(bd)} for block shape | exception | error | jax, pallas, tpu, block-shape, api-misuse |
| All block dimensions must be Elements or none of them can be | validation | error | jax, pallas, tpu, block-shape, element-window |
| Unsupported pipeline mode: {pipeline_mode}. | validation | error | jax, pallas, tpu, pipeline-mode, api-misuse |
| Lookahead is not supported for XLA pipeline emitter lowering | validation | error | jax, pallas, tpu, pipelining, lookahead |
| Only single (1) and double (2) buffering are supported. Got | validation | error | jax, pallas, tpu, pipelining, buffer-count |
| Cannot do int indexing on TPU | validation | error | jax, pallas, tpu, indexing |
| Loading from an accumulator is not supported. Use `matmul_po | validation | error | jax, pallas, tpu, accumulator, matmul |
| PRNG keys must be loaded from SMEM. Did you set the memory s | validation | error | jax, pallas, tpu, prng, random, smem |
| Indexing into a ()-shaped Ref not yet supported on TPU. | validation | error | jax, pallas, tpu, scalar-ref, smem |
| Can only load scalars from SMEM | validation | error | jax, pallas, tpu, smem, shape-mismatch |
| Loads are only allowed on VMEM and SMEM references. | validation | error | jax, pallas, tpu, memory-space, async-copy |
| Seed key_data must be 1D. | validation | error | jax, pallas, tpu, prng, shape-validation |
| Leading dimension of seed key_data must be 1. | validation | error | jax, pallas, tpu, prng, shape-validation |
| Can only load a single key per load. | validation | error | jax, pallas, tpu, prng, indexing |
| masked swap with non-32-bit data | validation | error | jax, pallas, tpu, masked-store, dtype |
| Expected value and mask to have the same shape, but got valu | validation | error | jax, pallas, tpu, shape-mismatch, masked-store |
| SMEM store does not support masks | validation | error | jax, pallas, tpu, smem, masked-store |
| Can only store scalars to SMEM | exception | error | jax, pallas, tpu, smem, shape-mismatch |
| Loads and stores are only allowed on VMEM and SMEM reference | exception | error | jax, pallas, tpu, memory-space, async-copy |
| Cannot store scalars to VMEM | exception | error | jax, pallas, tpu, vmem, scalar-store |
| Cannot swap scalars to VMEM. | exception | error | jax, pallas, tpu, swap, vmem, scalar |
| masked swap with strided store | exception | error | jax, pallas, tpu, strided-store, masked-store |
| Reductions over unsigned integers not implemented. | exception | error | jax, pallas, tpu, reduction, unsigned-integer, dtype |
| Reductions over {x_aval.dtype} not implemented. | exception | error | jax, pallas, tpu, reduction, dtype |
| `broadcast_to` is a Triton-specific primitive. Please consid | exception | error | jax, pallas, tpu, mosaic, broadcast |
| Unsigned integer dtype {aval.dtype} is not supported for dot | exception | error | jax, pallas, tpu, dot-general, unsigned-integer, dtype |
| {ctx.avals_out[0].dtype} | exception | error | jax, pallas, tpu, matmul, dtype, complex |
| Unsupported {preferred_element_type=} | exception | error | jax, pallas, tpu, dot-general, preferred-element-type, mixed-precision |
| Per-operand dot precision unsupported | exception | error | jax, pallas, tpu, dot-general, precision |
| Unsupported dot precision: {precision} | exception | error | jax, pallas, tpu, precision, dot-general |
| Per-operand conv precision unsupported | exception | error | jax, pallas, tpu, convolution, precision |
| Unsupported conv precision: {precision} | exception | error | jax, pallas, tpu, convolution, precision |
| Grouped convolutions are not supported on Pallas Mosaic TPU | exception | error | jax, pallas, tpu, convolution, grouped-conv, depthwise |
| Requires libtpu >= 0.1.0 | exception | error | jax, pallas, tpu, libtpu, version, convolution |
| Unsigned integer dtype {aval.dtype} is not supported for con | exception | error | jax, pallas, tpu, convolution, unsigned-integer, dtype |
| Unsupported cast: {from_dtype} -> {to_dtype} | exception | error | jax, pallas, tpu, dtype-cast, unsigned-integer |
| Only arrays with 32-bit element types can be converted to sc | exception | error | jax, pallas, tpu, squeeze, scalar, dtype |
| Dimension must be 0 for 1D iota. | exception | error | jax, pallas, tpu, iota, indexing |
| Only take_along_axis-like gathers supported | exception | error | jax, pallas, tpu, gather, not-implemented |
| Only gathers along the two minormost dimensions supported on | exception | error | jax, pallas, tpu, tensorcore, gather |
| Unsupported gather | exception | error | jax, pallas, tpu, gather, not-implemented |
| {aval_out.dtype} | exception | error | jax, pallas, tpu, dtype, add |
| Unsupported constant type: {x.type} | exception | error | jax, pallas, mlir, constant-folding, internal |
| Unsupported combination of input dtype ({x_aval.dtype}) and | exception | error | jax, pallas, tpu, argmax, dtype |
| Only single axis reduction supported | exception | error | jax, pallas, tpu, argmax, reduction |
| Pallas top_k only supports float32 and bfloat16, got {input_ | exception | error | jax, pallas, tpu, top-k, dtype |
| float32 top_k is not supported on TPUv3 or older | exception | error | jax, pallas, tpu, top-k, hardware-version |
| bfloat16 top_k is not supported on TPUv5 or older | exception | error | jax, pallas, tpu, top-k, hardware-version, bfloat16 |
| Unsupported dtype for sign: {x.dtype} | exception | error | jax, pallas, tpu, dtype, sign |
| Not implemented: accuracy | exception | error | jax, pallas, tpu, rsqrt, accuracy |
| Transforms not supported for matmul_acc_lhs. | exception | error | jax, pallas, tpu, matmul, not-implemented |
| Transforms not supported for matmul_pop. | exception | error | jax, pallas, tpu, matmul, not-implemented |
| Element with {padding=} is not supported. | validation | error | jax, pallas, mosaic, pipeline, blockspec, padding |
| Unsupported block dimension type: {type(bd)} | exception | error | jax, pallas, blockspec, type-validation |
| next_fetch is None | exception | error | jax, pallas, internal-invariant, prefetch |
| Grid must be specified when using lookahead. | exception | error | jax, pallas, pipeline, lookahead, grid |
| Output buffering does not support lookahead. | exception | error | jax, pallas, pipeline, lookahead, output-buffer |
| Cannot partition grid over dynamic number of cores. | exception | error | jax, pallas, tpu, grid-partition, static-value |
| dimension_semantics must be the same length as grid. | exception | error | jax, pallas, dimension-semantics, validation |
| Cannot partition over cores without parallel grid dimensions | exception | error | jax, pallas, tpu, grid-partition, dimension-semantics |
| Cannot partition over multiple dynamic parallel dimensions: | exception | error | jax, pallas, grid-partition, dynamic-shape |
| Only one of src or dst can be a BufferedRef. | exception | error | jax, pallas, pipeline, copy |
| One of src or dst must be a BufferedRef. | exception | error | jax, pallas, pipeline, copy |
| Grid must consist of Python integers and JAX Arrays: {grid_t | exception | error | jax, pallas, grid, validation |
| All elements in the grid must be strictly positive, but got | exception | error | jax, pallas, grid, validation |
| Only one of `core_axis` or `core_axis_name` can be set. | exception | error | jax, pallas, mutually-exclusive-args, validation |
| The emit_pipeline body function must return None. | exception | error | jax, pallas, kernel-body, validation |
| WriteEffect should not apply to an input buffer {ref_invar_i | exception | error | jax, pallas, effects, input-mutation |
| WriteEffect on input buffer {ref_invar_idx} | exception | error | jax, pallas, effects, input-mutation |
| Grid mapping with hijax index maps are not currently support | exception | error | jax, pallas, hijax, index-map, not-implemented |
| Batching over custom allocations is not supported yet. | exception | error | jax, pallas, vmap, batching, allocations, not-implemented |
| Batching over dynamic grid values is not supported yet. | exception | error | jax, pallas, mosaic, vmap, batching, tpu, not-implemented |
| Not implemented: bitcast 1D | exception | error | jax, pallas, mosaic, bitcast, dtype, shape-validation |
| Not implemented: the 2nd minor dim can not be perfectly pack | exception | error | jax, pallas, mosaic, bitcast, dtype, packing, shape-validation |
| shift must be non-negative. | exception | error | jax, pallas, mosaic, roll, argument-validation, shift |
| axis is out of range. | exception | error | jax, pallas, mosaic, roll, axis-validation, shape-validation |
| stride and stride_axis must be both specified or not. | exception | error | jax, pallas, mosaic, roll, argument-validation, kwargs-pairing |
| stride must be non-negative. | exception | error | jax, pallas, mosaic, roll, argument-validation, stride |
| stride_axis is out of range | exception | error | jax, pallas, mosaic, roll, axis-validation, shape-validation |
| expected axis and stride_axis are different. | exception | error | jax, pallas, mosaic, roll, argument-validation |
| Either both or neither `src_sem` and `device_id` can be set. | exception | error | jax, pallas, mosaic, dma, remote-copy, argument-validation |
| Cannot `wait_send` on a local copy. | exception | error | jax, pallas, mosaic, dma, remote-copy, wait-send |
| dma_start not implemented in LoJAX yet. | exception | error | jax, pallas, mosaic, lojax, dma, not-implemented, tpu |
| DMA start priority must be non-negative: {priority} | exception | error | jax, pallas, mosaic, dma, priority, argument-validation |
| DMA source/destination/semaphore arguments must be Refs. | exception | error | jax, pallas, mosaic, dma, ref-vs-array, type-validation |
| Cannot signal on a non-()-shaped semaphore: {dst_sem_shape} | exception | error | jax, pallas, mosaic, dma, semaphore, shape-validation |
| DMA source semaphore must be a Ref. | exception | error | jax, pallas, mosaic, dma, semaphore, ref-vs-array, type-validation |
| Cannot signal on a non-()-shaped semaphore: {src_sem_shape} | exception | error | jax, pallas, mosaic, dma, semaphore, shape-validation |
| DMA partial discharge add=True not yet implemented. | exception | error | jax, pallas, mosaic, dma, discharge, not-implemented, pipelining |
| The number of sources must match the packing factor ({packin | exception | error | jax, pallas, tpu, dtype, shape-mismatch |
| The bitwidth of `x` must match the bitwidth of `unpacked_dty | exception | error | jax, pallas, tpu, dtype, bitwidth |
| Index {index} is out of bounds for packing factor {packing_f | exception | error | jax, pallas, tpu, index-out-of-bounds |
| with_memory_space_constraint only supports HBM, VMEM, SMEM, | exception | error | jax, pallas, tpu, memory-space, unsupported-operation |
| trace_value requires a scalar value, got shape {value.shape} | exception | error | jax, pallas, tpu, debugging, scalar-required |
| trace_value requires i32 or f32, got {value.dtype} | exception | error | jax, pallas, tpu, debugging, dtype |
| load_staged_rhs must be an integer or None. | exception | error | jax, pallas, tpu, matmul, type-error |
| Expected an accumulator ref, got {acc} | exception | error | jax, pallas, tpu, matmul, memory-space |
| The shape of the accumulator {acc_shape} is not 2-dimensiona | exception | error | jax, pallas, tpu, matmul, shape-mismatch |
| The shape of the accumulator {acc_shape} does not match the | exception | error | jax, pallas, tpu, matmul, shape-mismatch |
| Only float32 and int32 results are supported, got {dtype} | exception | error | jax, pallas, tpu, matmul, dtype |
| Bit width must be 32 | exception | error | jax, pallas, tpu, random, prng |
| Cannot split a Pallas key. Use fold_in instead to generate n | exception | error | jax, pallas, tpu, random, prng, not-implemented |
| block_size ({len(block_size)}) and tile_size ({len(tile_size | exception | error | jax, pallas, tpu, random, shape-mismatch |
| The current TPU does not have SparseCores | exception | critical | jax, pallas, tpu, sparsecore, hardware, environment |
| Mesh has {self.num_cores} cores, but the current TPU chip ha | exception | error | jax, pallas, tpu, sparsecore, hardware, resource-limits |
| You can't use two different ScalarSubcoreMeshes. | exception | error | jax, pallas, tpu, sparsecore, mesh-compat |
| {self} should have the same core axis name and number of cor | validation | error | jax, pallas, tpu, sparsecore, mesh-compat |
| {self} should have a different axis name from the TensorCore | validation | error | jax, pallas, tpu, sparsecore, axis-naming |
| Mesh has {self.num_subcores} subcores, but the current TPU c | validation | error | jax, pallas, tpu, hardware-mismatch, sparsecore |
| You can't use two different VectorSubcoreMeshes. | validation | error | jax, pallas, tpu, sparsecore, mesh-compat |
| {self} should have the same core axis name and number of cor | validation | error | jax, pallas, tpu, sparsecore, mesh-compat |
| {self} should have a different core axis name from the Tenso | validation | error | jax, pallas, tpu, sparsecore, axis-naming |
| {self} should have a different subcore axis name from the Te | validation | error | jax, pallas, tpu, sparsecore, axis-naming |
| Unsupported dtype: {dtype} | validation | error | jax, pallas, tpu, sparsecore, dtype |
| Get does not support loading from {ref_memory_space!r}. Copy | exception | error | jax, pallas, tpu, sparsecore, memory-space |
| Integer indexing of refs that follows a non-trivial slice is | exception | error | jax, pallas, tpu, sparsecore, indexing |
| Get only supports slices with stride 1, got {strides} | exception | error | jax, pallas, tpu, sparsecore, slicing, strides |
| Get only supports loading scalars from SMEM. | exception | error | jax, pallas, tpu, sparsecore, scalar-load |
| Get does not support masked scalar loads | exception | error | jax, pallas, tpu, sparsecore, masked-load |
| Swap does not support storing to {ref_memory_space!r}. Copy | exception | error | jax, pallas, tpu, sparsecore, memory-space, swap |
| Swap only supports slices with stride 1, got {strides} | exception | error | jax, pallas, tpu, sparsecore, swap, memory-space |
| Swap only supports scalars in SMEM. | exception | error | jax, pallas, tpu, sparsecore, masked-store, swap |
| Swap does not support masked scalar stores | exception | error | jax, pallas, tpu, sparsecore, atomic, swap |
| Gather only supports loading from VMEM, got {ref_aval.memory | exception | error | jax, pallas, tpu, sparsecore, memory-space, gather |
| {x.shape=} does not match expected shape {expected_shape} | exception | error | jax, pallas, sparsecore, scatter, shape-mismatch |
| val.dtype={x.dtype} != ref.dtype={ref.dtype} | exception | error | jax, pallas, sparsecore, scatter, dtype-mismatch |
| {mask.shape=} does not match expected shape {expected_shape} | exception | error | jax, pallas, sparsecore, scatter, mask-shape |
| Scatter only supports VectorSubcoreMesh, got {type(ref_aval. | exception | error | jax, pallas, sparsecore, mesh, scatter |
| Scatter only supports storing to VMEM, got {memory_space} | exception | error | jax, pallas, sparsecore, scatter, memory-space |
| Indices must not be empty | exception | error | jax, pallas, sparsecore, scatter, empty-indices |
| Cannot bitcast a ()-shaped array to a dtype with a different | exception | error | jax, pallas, sparsecore, bitcast, scalar |
| Cannot bitcast from {x.dtype} ({old_bitwidth} bits) to {dtyp | exception | error | jax, pallas, sparsecore, bitcast, divisibility |
| x.dtype={x.dtype} must be uint32, int32 or float32 | exception | error | jax, pallas, sparsecore, scan-count, dtype |
| mask.dtype={mask.dtype} is not a boolean dtype | exception | error | jax, pallas, sparsecore, scan-count, mask-dtype |
| x.shape={x.shape} != mask.shape={mask.shape} | exception | error | jax, pallas, sparsecore, scan-count, shape-mismatch |
| reductions require axes to be (0,) on SparseCore, but got {a | exception | error | jax, pallas, sparsecore, reduction, axes |
| cummax: x={x.aval} must be rank 1 | exception | error | jax, pallas, sparsecore, cummax, rank |
| packed cannot be specified if layout is specified. | validation | error | jax, pallas, mosaic-gpu, tmem, layout, api-misuse |
| packed, collective and layout arguments are only supported f | validation | error | jax, pallas, mosaic-gpu, smem, gmem, tmem, api-misuse |
| Cannot specify both out_shape and out_type. Use out_type. | exception | error | jax, pallas, mosaic-gpu, deprecation, api-migration |
| Cannot specify both scratch_shapes and scratch_types. Use sc | exception | error | jax, pallas, mosaic-gpu, deprecation, scratch-memory |
| Unsupported dtype: {ref.dtype} | exception | error | jax, pallas, mosaic-gpu, dtype, shared-memory, not-implemented |
| Only byte-aligned shapes are supported. Got shape: {ref.dtyp | exception | error | jax, pallas, mosaic-gpu, shared-memory, alignment, shape-validation |
| Only SMEM and TMEM refs are supported. | exception | error | jax, pallas, mosaic-gpu, ref-union, memory-space, not-implemented |
| Ref unions can't be assigned to. | exception | error | jax, pallas, mosaic-gpu, ref-union, immutable, read-only |
| layout attribute is only defined for TMEM refs | exception | error | jax, pallas, mosaic-gpu, tmem, gpu-kernels |
| collective attribute is only defined for TMEM refs | exception | error | jax, pallas, mosaic-gpu, tmem, collective-ops |
| Some aliased TMEM references are collective and some are not | exception | error | jax, pallas, mosaic-gpu, tmem, aliasing |
| All aliased Refs must have the same memory space (SMEM or TM | exception | error | jax, pallas, mosaic-gpu, smem, tmem, aliasing |
| Cannot transform type: {x} | exception | error | jax, pallas, abstract-values, transforms, type-dispatch |
| Transpose cannot be moved before a tiling transform when it | exception | error | jax, pallas, mosaic-gpu, transpose, tiling, transforms |
| Strided slices unsupported. Got stride: {ds.stride} | exception | error | jax, pallas, slicing, strides, tiling |
| Expected slice start ({start}) and slice size ({size}) to be | exception | error | jax, pallas, alignment, tiling, slicing |
| Unsupported index type: {type(idx)} | exception | error | jax, pallas, indexing, type-dispatch, tiling |
| Commuting a `UntilingTransform` with a `ReshapeTransform` is | exception | error | jax, pallas, mosaic-gpu, reshape, transforms |
| Commuting a `UntilingTransform` with a `ReshapeTransform` is | exception | error | jax, pallas, mosaic-gpu, tiling, transforms |
| Unfolding dimensions is not supported when commuting an `Un | exception | error | jax, pallas, mosaic-gpu, reshape, unfolding |
| Folding tiled dimensions into untiled dimensions is not supp | exception | error | jax, pallas, mosaic-gpu, reshape, tiling |
| Cannot commute `UntilingTransform` with `ReshapeTransform` w | exception | error | jax, pallas, mosaic-gpu, reshape, tiling, layout |
| Unsupported transform: {type(transform)} | exception | error | jax, pallas, mosaic-gpu, transforms, not-implemented |
| ref must be a reference | exception | error | jax, pallas, mosaic-gpu, peer-memory, type-error |
| Can't make a multicast reference into a peer reference. | exception | error | jax, pallas, mosaic-gpu, peer-memory, multicast |
| dims and idxs must have the same length | exception | error | jax, pallas, mosaic-gpu, cluster, validation |
| Can't make a peer reference into a multicast reference. | exception | error | jax, pallas, mosaic-gpu, multicast, peer-memory |
| Can't transpose a TMEM reference. | exception | error | jax, pallas, mosaic-gpu, tmem, transpose |
| Unsupported type: {x} | exception | error | jax, pallas, mosaic-gpu, aval, type-error |
| Swizzle {self.swizzle} is not supported. Only 32, 64 and 128 | exception | error | jax, pallas, mosaic-gpu, swizzle, validation |
| Swizzle {self.swizzle} requires the trailing dimension to be | exception | error | jax, pallas, mosaic-gpu, swizzle, layout, shape-mismatch |
| Can't transpose the swizzled dimension. | exception | error | jax, pallas, mosaic-gpu, transpose, swizzle |
| Reshape shape {shape} is not divisible by swizzle elements { | validation | error | jax, pallas, reshape, swizzle, mosaic-gpu |
| Cannot commute unswizzle and indexer with {aval}, which does | validation | error | jax, pallas, dtype, swizzle, indexing |
| Non-slice indices are not supported in 2 minormost dims: {id | validation | error | jax, pallas, indexing, swizzle, not-implemented |
| Swizzled dims cannot be sliced | validation | error | jax, pallas, slicing, swizzle |
| Unsupported ndim: {x.ndim} | validation | error | jax, pallas, ndim, shape-validation |
| Unsupported shape: {x.shape} | validation | error | jax, pallas, shape-validation, ndim |
| Second dimension {x.shape[1]} must be divisible by batch_siz | validation | error | jax, pallas, shape-validation, divisibility |
| Expected indexer to have exactly {k + 2} dimensions, but got | validation | error | jax, pallas, indexing, rank-mismatch |
| Slicing batch dimensions is not supported. | validation | error | jax, pallas, indexing, slicing, not-implemented |
| collective_axes is not supported in pallas_call. Use plgpu.k | validation | error | jax, pallas, mosaic-gpu, collective-axes, unsupported-feature |
| Num arrivals must be at least 1, but got {n} | validation | error | jax, pallas, barrier, validation, config |
| Barriers are not arrays | validation | error | jax, pallas, barrier, api-misuse |
| Cluster barriers are not arrays | validation | error | jax, pallas, barrier, cluster, api-misuse |
| Preinitialized WGMMAAccumulatorRef only supported in pl.run_ | validation | error | jax, pallas, wgmma, accumulator, api-misuse |
| Non-trivial indexing on WGMMAAbstractAccumulatorRef is not s | validation | error | jax, pallas, mosaic-gpu, wgmma, not-implemented, indexing |
| cluster= must be at most 3D, got {self}. | validation | error | jax, pallas, mosaic-gpu, cluster, launch-config, validation |
| grid_names must have the same length as grid, got {self}. | validation | error | jax, pallas, mosaic-gpu, launch-config, validation, shape-mismatch |
| cluster_names must have the same length as cluster, got {sel | validation | error | jax, pallas, mosaic-gpu, launch-config, validation, shape-mismatch |
| num_threads and thread_name must be either both set or both | validation | error | jax, pallas, mosaic-gpu, warp-specialization, validation |
| Requested too many CUDA threads per block. Each Mosaic threa | validation | error | jax, pallas, mosaic-gpu, thread-limits, resource-limit, validation |
| Can't instantiate {self} with arguments. | validation | error | jax, pallas, mosaic-gpu, layout, api-misuse, validation |
| Only TiledLayout supports reductions. | validation | error | jax, pallas, mosaic-gpu, layout, reduction, unsupported-type |
| Invalid TMEMLayout: {self} | validation | error | jax, pallas, mosaic-gpu, tmem-layout, enum, version-mismatch |
| {tiling=} and {grid=} must have same length. | validation | error | jax, pallas, mosaic-gpu, nd-loop, tiling, shape-mismatch |
| Tiling {tiling} does not divide grid {grid}. | validation | error | jax, pallas, mosaic-gpu, nd-loop, tiling, divisibility |
| Expected metadata dtype to be uint2, got: {meta.dtype} | validation | error | jax, pallas, mosaic-gpu, sparse, tcgen05, dtype, validation |
| Expected metadata to be 3-dimensional (M, K // 4, 2), but it | validation | error | jax, pallas, mosaic-gpu, sparse, tcgen05, shape, rank |
| Expected the trailing dimension of the metadata to be 2, got | validation | error | jax, pallas, mosaic-gpu, sparse, tcgen05, shape, last-dim |
| Sparse metadata format not implemented for {operand_dtype=} | validation | error | jax, pallas, mosaic-gpu, sparse, tcgen05, dtype, not-implemented |
| No valid out swizzle{what}: minor dimension has {minor_dim_b | validation | error | jax, pallas, mosaic-gpu, swizzle, shared-memory, alignment, matmul |
| `thread_id` must be zero when allocating a buffer for all th | exception | error | jax, pallas, mosaic-gpu, interpret-mode, allocation, warp-specialization |
| `block_id` must be zero when allocating a buffer for all thr | exception | error | jax, pallas, mosaic-gpu, interpret-mode, allocation, block-id |
| Out-of-bounds read of {allocation_key}: reading [{read_range | exception | error | pallas, mosaic-gpu, interpret-mode, out-of-bounds, jax |
| Out-of-bounds block index {block_indices} for {allocation_ke | exception | error | pallas, mosaic-gpu, interpret-mode, out-of-bounds, input-block |
| Dynamic indexing not supported in GPU interpret mode | exception | error | pallas, mosaic-gpu, interpret-mode, dynamic-indexing |
| Unsupported transform: {transform} | exception | error | pallas, mosaic-gpu, interpret-mode, unsupported-transform, internal |
| Out-of-bounds swap of {allocation_key}: swapping [{read_writ | exception | error | pallas, mosaic-gpu, interpret-mode, out-of-bounds, atomics, swap |
| Out-of-bounds masked swap of {allocation_key}: swapping [{re | exception | error | pallas, mosaic-gpu, interpret-mode, out-of-bounds, atomics, mask |
| Unsupported barrier type: {type(barrier)} | exception | error | pallas, mosaic-gpu, interpret-mode, barrier, unsupported |
| predicate not supported | exception | error | pallas, mosaic-gpu, interpret-mode, predicate, not-implemented |
| reduction_op not supported | exception | error | pallas, mosaic-gpu, interpret-mode, reduction, not-implemented |
| tcgen05_mma only allows arriving on a Barrier | exception | error | pallas, mosaic-gpu, tcgen05, mma, barrier, interpret-mode |
| tcgen05_mma only allows arriving on a Barrier that orders te | exception | error | pallas, mosaic-gpu, tcgen05, mma, barrier, tensor-core |
| tcgen05_commit_arrive only allows arriving on a Barrier | exception | error | pallas, mosaic-gpu, tcgen05, barrier, interpret-mode |
| tcgen05_commit_arrive only allows arriving on a Barrier that | exception | error | pallas, mosaic-gpu, tcgen05, barrier, tensor-core |
| Dynamic grid bounds not (yet) supported in GPU interpret mod | exception | error | pallas, mosaic-gpu, interpret-mode, dynamic-grid, not-implemented |
| Unsupported mesh type: {type(mesh)} | exception | error | pallas, mosaic-gpu, mesh, interpret-mode, type-validation |
| Invalid grid {grid_mapping.grid} in grid_mapping: expected g | exception | error | pallas, mosaic-gpu, grid, mesh, config-mismatch |
| Dynamic grid bounds not (yet) supported on GPU | exception | error | pallas, mosaic-gpu, interpret-mode, dynamic-grid, not-implemented |
| Scalar arguments not (yet) supported on GPU | exception | error | pallas, mosaic-gpu, interpret-mode, scalar-args, not-implemented |
| Unsupported memory space: {space} | exception | error | pallas, mosaic-gpu, memory-space, interpret-mode, not-implemented |
| Requesting collective allocations, but no explicit thread ax | exception | error | pallas, mosaic-gpu, collective, mesh, thread-axis, config |
| Collective allocations along cluster axes are not supported. | exception | error | jax, mosaic-gpu, interpret-mode, collective-allocations |
| Scoped allocation must have the thread axis in its collectiv | exception | error | jax, mosaic-gpu, run-scoped, collective-axes |
| Indexing barrier with {transforms} not supported in GPU inte | exception | error | jax, mosaic-gpu, barrier, transforms |
| Expected an `NDIndexer`, but got {transforms[0]} | exception | error | jax, mosaic-gpu, barrier, type-validation |
| Attempting to operate on barrier without indexing, but `num_ | exception | error | jax, mosaic-gpu, barrier, indexing |
| Dynamic grid bounds not supported in the Mosaic GPU lowering | exception | error | |
| Data type bitcast is only supported from i8 to other types. | exception | error | jax, pallas, mosaic-gpu, bitcast, dtype, not-implemented |
| Only workgroup memory is supported but got {ref}. | exception | error | jax, pallas, mosaic-gpu, smem, memory-space, bitcast |
| Data type bitcast is only supported for contiguous 1D arrays | exception | error | jax, pallas, mosaic-gpu, bitcast, stride, contiguity |
| Can not bitcast memory region of size {shape_bitwidth} bits | exception | error | jax, pallas, mosaic-gpu, bitcast, size-mismatch, alignment |
| TMEM aliasing only supported for Refs with the same first di | exception | error | jax, pallas, tmem, tcgen05, aliasing, shape-mismatch |
| Only byte-aligned bitcasts are supported. | exception | error | jax, pallas, mosaic-gpu, smem, bitcast, alignment, sub-byte |
| The base ref for aliases must come from a slice_smem op. | exception | error | jax, pallas, mosaic-gpu, warpgroup, smem, aliasing, not-implemented |
| Unsupported TMEM ref {ref}. | exception | error | jax, pallas, tmem, tcgen05, aliasing, not-implemented |
| Unsupported memory space. | exception | error | jax, pallas, mosaic-gpu, memory-space, aliasing, not-implemented |
| Tiling without swizzle is not supported. | exception | error | |
| Transforms are not yet implemented for warpgroup semantics | exception | error | jax, pallas, mosaic-gpu, transforms, warpgroup |
| Stores to TMEM are asynchronous operations and cannot be per | exception | error | jax, pallas, mosaic-gpu, tmem, blackwell, tcgen05 |
| Can only store scalars in warp-level lowering. | exception | error | jax, pallas, mosaic-gpu, warp-semantics, store |
| Can only store to references (got {x_ref}). | exception | error | jax, pallas, mosaic-gpu, type-error, internal-api |
| Can only store scalars or vectors (got {value}). | exception | error | jax, pallas, mosaic-gpu, type-error, warpgroup |
| Can only store to references (got {x_smem}). | exception | error | jax, pallas, mosaic-gpu, type-error, warpgroup |
| out_dtype argument in binary_op_lowering_rule_wg | exception | error | jax, pallas, mosaic-gpu, dtype, lowering, not-implemented |
| {ctx.prim} does not support {x_aval.dtype} and {y_aval.dtype | exception | error | jax, pallas, mosaic-gpu, dtype, binary-op, not-implemented |
| Unsupported dtype {x_aval.dtype} | exception | error | jax, pallas, mosaic-gpu, integer-pow, complex-dtype, not-implemented |
| Not implemented: accuracy | exception | error | jax, pallas, mosaic-gpu, rsqrt, accuracy, not-implemented |
| No axis names are available. Make sure you are using `pl.cor | exception | error | jax, pallas, mosaic-gpu, cluster, mesh, core-map |
| Unknown cluster axis {axis_name}, available axes: {[*axis_na | exception | error | jax, pallas, mosaic-gpu, cluster, axis-name, mesh |
| No axis names are available. Make sure you are using `pl.cor | exception | error | jax, pallas, mesh, collective, axis-name |
| Axis {axis_name} does not refer to a GPU mesh axis (availabl | exception | error | jax, pallas, mesh, collective, axis-name |
| Ordered debug_print is not supported on Pallas. | exception | error | jax, pallas, debug-print, ordered |
| Only positional arguments are supported by debug_print on Pa | exception | error | jax, pallas, debug-print, kwargs |
| Can only print scalars in warp-level code. | exception | error | jax, pallas, debug-print, warp-level |
| debug_print only supports printing of scalar values, or a si | exception | error | jax, pallas, debug-print, arrays |
| Buffer poisoning is not supported on GPU yet. | exception | error | jax, pallas, buffer-poisoning, debug-flag, gpu |
| Only thread-collective allocations are supported in run_scop | exception | error | jax, pallas, run-scoped, collective-axes, warp-specialization |
| WGMMA accumulators can only be allocated non-collectively. H | exception | error | jax, pallas, wgmma, run-scoped, collective-axes |
| Invalid WGMMA accumulator dtype for s8/i8 WGMMA. Expected si | exception | error | jax, pallas, wgmma, int8, dtype |
| Only thread-collective allocations are supported in multithr | exception | error | jax, pallas, run-scoped, warp-specialization, collective-axes |
| Can't convert to ref: {aval} | exception | error | jax, pallas, run-scoped, ref, type-error |
| get_global only supports semaphores, got {what} | exception | error | jax, pallas, get-global, semaphore |
| Uninitialized Refs are not supported in lowering of run_stat | exception | error | jax, pallas, run-state, uninitialized-ref |
| Integer constant out of range for i64: {v} | exception | error | pallas, mosaic-gpu, integer-overflow, jax |
| Unsupported index: {v} of type {type(v)} | exception | error | pallas, mosaic-gpu, indexing, type-error |
| Cannot merge empty list of indexers | exception | error | pallas, internal-api, indexing |
| Non-unit strides not implemented. | exception | error | pallas, mosaic-gpu, slicing, not-implemented |
| Unhandled transforms for semaphore_read: {transforms} | exception | error | pallas, mosaic-gpu, semaphore, not-implemented |
| Non-debug checks are not supported by the Mosaic GPU backend | exception | error | pallas, checkify, mosaic-gpu, not-implemented |
| relayout_p is not supported with Lane semantics. | exception | error | pallas, mosaic-gpu, layout, not-implemented |
| Only plgpu.Layout.WG_SPLAT is supported for scalar values. | exception | error | pallas, mosaic-gpu, layout, scalar |
| Failed to infer the output layout of the iota. Please apply | exception | error | pallas, mosaic-gpu, iota, layout |
| Encountered non-leading UntilingTransform or UnswizzleRef tr | exception | error | pallas, mosaic-gpu, transforms, not-implemented |
| dynamic grid bounds not supported in the Mosaic GPU backend | exception | error | pallas, mosaic-gpu, grid, not-implemented |
| Mesh {mesh} is not supported by the Mosaic GPU backend | exception | error | pallas, mosaic-gpu, mesh, not-implemented |
| Lowering multiple mesh/function pairs is not supported by th | exception | error | pallas, mosaic-gpu, mpmd-map, not-implemented |
| External meshes are not supported by the Mosaic GPU backend | exception | error | pallas, mosaic-gpu, mpmd-map, mesh, not-implemented |
| Scratch operands are not supported by the Mosaic GPU mpmd_ma | exception | error | pallas, mosaic-gpu, mpmd-map, scratch, not-implemented |
| Unsupported block size type: {type(bd)} | exception | error | pallas, mosaic-gpu, blockspec, not-implemented |
| Only collective_axes that include all JAX device mesh axes | exception | error | jax, pallas, mosaic-gpu, collective-axes, device-mesh, multicast |
| Non-indexing transforms on GMEM refs are not implemented. | exception | error | jax, pallas, mosaic-gpu, gmem-ref, transforms, not-implemented |
| The stored value has shape {src.shape}, but the target refer | exception | error | jax, pallas, mosaic-gpu, shape-mismatch, async-store |
| The stored value has dtype {src.dtype}, but the target refer | exception | error | jax, pallas, mosaic-gpu, dtype-mismatch, async-store |
| Expected a single barrier, got a barrier reference with shap | exception | error | jax, pallas, mosaic-gpu, barrier, async-store |
| Scalars are not supported in async_store_smem | exception | error | jax, pallas, mosaic-gpu, scalar, async-store, not-implemented |
| Unexpected unhandled transforms: {remaining_ref_transforms} | exception | error | jax, pallas, mosaic-gpu, internal-invariant, transforms, async-store |
| async_store_smem requires a tiled and swizzled ref | exception | error | jax, pallas, mosaic-gpu, tiled-layout, swizzle, async-store |
| Can only transfer integer bytes (shape={shape}, dtype={dtype | exception | error | jax, pallas, mosaic-gpu, sub-byte-dtype, alignment, async-store |
| Transfer is not a multiple of {WARPGROUP_SIZE} bytes | exception | error | jax, pallas, mosaic-gpu, alignment, warpgroup, async-store |
| Expected source shape to be {expected_src_shape}, but got {s | exception | error | jax, pallas, mosaic-gpu, shape-mismatch, sharding, gmem-copy |
| Can only transfer integer bytes (shape={dst_ty.shape}, dtype | exception | error | jax, pallas, mosaic-gpu, sub-byte-dtype, alignment, gmem-copy |
| Expected exactly one collective axis, got {collective_axes=} | exception | error | jax, pallas, mosaic-gpu, collective-axes, leader-tracked |
| Partitioned loads only supported for clusters of size 2. Got | exception | error | jax, pallas, mosaic-gpu, cluster-size, leader-tracked, not-implemented |
| copy_gmem_to_smem with a barrier is only supported Hopper an | exception | error | jax, pallas, mosaic-gpu, gpu-architecture, tma, barrier, cp-async |
| Only the TMA implementation supports collective copies | exception | error | jax, pallas, mosaic-gpu, gpu-architecture, collective-axes, cp-async |
| Only the TMA implementation supports leader_tracked copies | exception | error | jax, pallas, mosaic-gpu, gpu-architecture, leader-tracked, cp-async |
| The cp.async implementation only supports oob_mode=OOBFillMo | exception | error | jax, pallas, mosaic-gpu, gpu-architecture, out-of-bounds, cp-async |
| The cp.async implementation does not support user-defined pr | exception | error | jax, pallas, mosaic-gpu, gpu-architecture, predicate, cp-async, not-implemented |
| copy_gmem_to_smem without a barrier is only supported on pre | exception | error | mosaic-gpu, pallas, gpu, barrier, tma, hopper |
| Only copies transferring a number of bytes divisible by the | exception | error | mosaic-gpu, pallas, alignment, warpgroup, copy |
| predicate is not supported with Warpgroup lowering in jaxlib | exception | error | mosaic-gpu, pallas, jaxlib-version, predicate, tma |
| Loading from a remote ref is only supported in jaxlib versio | exception | error | mosaic-gpu, pallas, jaxlib-version, multi-gpu, peer-id |
| `collective_axes` must be specified when `leader_tracked` is | validation | error | mosaic-gpu, pallas, api-misuse, collectives |
| wait_gmem_to_smem is only supported on pre-Hopper GPUs, whic | validation | error | mosaic-gpu, pallas, gpu, hopper, synchronization |
| GMEM refs with peer ids are not supported in warpgroup lower | validation | error | mosaic-gpu, pallas, prefetch, multi-gpu, peer-id |
| Barrier does not support slice with `stride != 1` | validation | error | mosaic-gpu, pallas, barrier, indexing, slice |
| Barrier can only be indexed with integers or slices, got {id | validation | error | mosaic-gpu, pallas, barrier, indexing, type-error |
| Barrier does not support arbitrary transforms | validation | error | mosaic-gpu, pallas, barrier, transforms |
| Arriving on a collective barrier is not supported in a warp | validation | error | mosaic-gpu, pallas, barrier, warp-specialization |
| Incompatible shapes for matrix multiplication: lhs={a.shape} | validation | error | mosaic-gpu, pallas, wgmma, shape-mismatch, matrix-multiply |
| Expected WGMMAAbstractAccumulatorRef got {acc_aval} | validation | error | mosaic-gpu, pallas, wgmma, accumulator, type-error |
| Accumulator shape {inner.shape} does not match value shape { | validation | error | jax, pallas, mosaic-gpu, wgmma, shape-mismatch |
| Accumulator dtype {inner.dtype} does not match value dtype { | validation | error | jax, pallas, mosaic-gpu, wgmma, dtype-mismatch |
| Accumulator and LHS have incompatible shapes. Expected LHS t | validation | error | jax, pallas, tcgen05, mma, shape-mismatch |
| Accumulator and RHS have incompatible shapes. Expected RHS t | validation | error | jax, pallas, tcgen05, collective-mma, shape-mismatch |
| Accumulator and RHS have incompatible shapes. Expected RHS t | validation | error | jax, pallas, tcgen05, mma, shape-mismatch |
| a_scale and b_scale must both be present or absent. | validation | error | jax, pallas, tcgen05, block-scaling, argument-validation |
| Accumulator must be a TMEM Ref. | validation | error | jax, pallas, tcgen05, tmem, memory-space |
| LHS must be a TMEM/SMEM Ref. | validation | error | jax, pallas, tcgen05, smem, tmem, memory-space |
| RHS must be an SMEM Ref. | validation | error | jax, pallas, tcgen05, smem, memory-space |
| Accumulator Ref must be collective if collective_axis is set | validation | error | jax, pallas, tcgen05, collective-mma, tmem |
| LHS Ref must be collective if collective_axis is set. | validation | error | jax, pallas, tcgen05, collective-mma, tmem |
| MMA barrier must have orders_tensor_core set to True. | validation | error | jax, pallas, tcgen05, barrier, pipelining |
| a_scale must be a TMEM Ref | validation | error | jax, pallas, tcgen05, block-scaling, tmem |
| b_scale must be a TMEM Ref | validation | error | jax, pallas, tcgen05, block-scaling, tmem |
| Unsupported transforms for ACC: {acc_transforms}. | exception | error | jax, pallas, tcgen05, transform, not-implemented |
| Unsupported transforms for LHS: {a_transforms}. | exception | error | jax, pallas, tcgen05, transform, not-implemented |
| MMA lhs tiling does not fit swizzle. {lhs_tiling=} expected= | validation | error | jax, pallas, tcgen05, swizzle, tiling, layout |
| Unsupported transforms for RHS: {b_transforms}. | exception | error | jax, pallas, tcgen05, transform, not-implemented |
| MMA rhs tiling does not fit swizzle {rhs_tiling=} expected={ | validation | error | jax, pallas, tcgen05, swizzle, tiling, layout |
| MMA rhs swizzle must match lhs swizzle. {lhs_swizzle=} {rhs_ | validation | error | jax, pallas, mosaic-gpu, tcgen05, mma, swizzle |
| TMEM transpose not allowed. | validation | error | jax, pallas, tcgen05, tmem, transpose |
| Unsupported transforms: {a_scale_transforms} | exception | error | jax, pallas, tcgen05, fp8, scaling, not-implemented |
| Unsupported transforms: {b_scale_transforms} | exception | error | jax, pallas, tcgen05, fp8, scaling, not-implemented |
| Unsupported transforms: {a_sparse_metadata_transforms} | exception | error | jax, pallas, tcgen05, sparse, not-implemented |
| Unsupported transforms for {ref}. Transforms {transforms}. | exception | error | jax, pallas, tcgen05, warp-group, transforms, not-implemented |
| No cluster axes found. | validation | error | jax, pallas, tcgen05, cluster, collective-mma |
| Can only perform collective MMA along minormost cluster axis | validation | error | jax, pallas, tcgen05, cluster, collective-mma, axis-order |
| inline_mgpu_p only supports plgpu.ShapeDtypeStruct return ty | validation | error | jax, pallas, mosaic-gpu, inline-mgpu, type-validation |
| inline_mgpu_p only supports only SomeLayout and RefType arg | validation | error | jax, pallas, mosaic-gpu, inline-mgpu, type-validation |
| Mismatched type shape: {treedef} != {treedef_ty} | validation | error | jax, pallas, inline-mgpu, pytree, signature-mismatch |
| Mismatched type: {a, t} | validation | error | jax, pallas, inline-mgpu, type-mismatch |
| inline_mgpu_p does not support discharge. | exception | error | jax, pallas, inline-mgpu, discharge, not-implemented |
| Array dtype mismatch: expected {v.mlir_dtype} got {mlir_dtyp | validation | error | jax, pallas, inline-mgpu, dtype-mismatch |
| Array shape mismatch: expected {ty.shape} got {v.shape}. | validation | error | jax, pallas, inline-mgpu, shape-mismatch |
| Array layout mismatch: expected {v.layout} got {ty.layout.to | validation | error | jax, pallas, inline-mgpu, layout-mismatch |
| Unexpected layout for {v} (expected: {ty}) | validation | error | jax, pallas, inline-mgpu, layout-mismatch |
| Unexpected type {ty} for value {v} | validation | error | jax, pallas, inline-mgpu, type-validation |
| inline_mgpu in a single-warp context only supports scalar ar | validation | error | jax, pallas, inline-mgpu, warp-semantics, scalar-required |
| Not all transforms could be handled. Remaining transforms: { | exception | error | jax, pallas, mosaic-gpu, transforms, not-implemented |
| Transform mismatch: got {user_transforms}, expected {ty_tran | validation | error | jax, pallas, transform-mismatch, validation |
| inline_mgpu in a single-warp context only supports scalar re | validation | error | jax, pallas, warp-semantics, scalar-return |
| inline_mgpu_p return type tree mismatch: {ret} != {return_ty | validation | error | jax, pallas, pytree, return-type-mismatch |
| Unsupported aval type: {aval}, {type(aval)}, {t} | validation | error | jax, pallas, aval, not-implemented, mlir |
| Expected a ShapeDtypeStruct, but got: {r} | validation | error | jax, pallas, shapedtypestruct, return-type |
| Expected a FragmentedArray, but got: {fa} | validation | error | jax, pallas, fragmented-array, lowering |
| Expected {result_shape} but got {fa.shape} | validation | error | jax, pallas, shape-mismatch, lowering |
| Output layout {out_layout} does not match the layout of the | validation | error | jax, pallas, layout-mismatch, mosaic-gpu |
| Expected 0D shape, but got {fa.shape} | validation | error | jax, pallas, scalar, shape-mismatch |
| Expected WGSplatFragLayout, but got {fa.layout} | validation | error | jax, pallas, layout, scalar, mosaic-gpu |
| Async load only supports TMEM refs | validation | error | jax, pallas, tmem, memory-space, blackwell |
| Unsupported dtype for reduction: {val_aval.dtype}. Only floa | validation | error | jax, pallas, tmem, dtype, reduction |
| Unimplemented transforms for TMEM refs. {transforms=} | validation | error | jax, pallas, tmem, transforms, not-implemented |
| Unsupported load reduce operation: {orig_reduce} | validation | error | jax, pallas, tmem, reduction, invalid-argument |
| Async store only supports TMEM refs | validation | error | jax, pallas, tmem, memory-space, blackwell |
| Only collective_axes that include all JAX device mesh ({mesh | exception | error | jax, pallas, tpu, distributed, device-mesh, multimem |
| Unhandled transforms for multimem_store: {transforms} | exception | error | jax, pallas, transforms, multimem, lowering |
| JAX device mesh is required by multimem_load_reduce, but not | exception | error | jax, pallas, device-mesh, multimem, missing-context |
| Failed to infer the output layout of multimem_load_reduce. P | exception | error | jax, pallas, layout, multimem, tpu |
| Only tiled and WG strided layouts are supported by multimem_ | validation | error | jax, pallas, layout, multimem, unsupported-layout |
| Unhandled transforms for multimem_load_reduce: {transforms} | exception | error | jax, pallas, transforms, multimem, lowering |
| Cannot pass the same ref into a mpmd map multiple times | exception | error | jax, pallas, mpmd, ref-aliasing, notimplementederror |
| MPMD map with more than one mesh requires scratch_type to ha | exception | error | jax, pallas, mpmd, tpu, scratch-memory, multi-mesh |
| The kernel function in mpmd_map {debug_info.func_src_info} s | exception | error | jax, pallas, mpmd, kernel-api, inplace-semantics |
| input pinned buffers without input_output_aliases:{missing} | exception | error | jax, pallas, pallas-call, input-output-alias, pinned-buffer |
| pallas_call requires all mesh axes to be Manual, got {get_ab | exception | error | jax, pallas, pallas-call, sharding, mesh, manual-axes |
| pallas_call does not support hijax for index_map | exception | error | jax, pallas, pallas-call, index-map, lowering, notimplementederror |
| interpret with dynamic grid bounds unsupported | exception | error | jax, pallas, jvp, autodiff, dynamic-grid, notimplementederror |
| JVP with aliasing not supported. | exception | error | jax, pallas, jvp, autodiff, input-output-alias, notimplementederror |
| pallas_call with a mesh does not support JVP | exception | error | jax, pallas, jvp, autodiff, mesh, spmd, notimplementederror |
| vmapping pallas_call with no arguments. | exception | error | jax, pallas, vmap, batching, not-implemented |
| pallas_call with a mesh does not support batching | exception | error | jax, pallas, vmap, mesh, sharding, not-implemented |
| The kernel function in the pallas_call {debug_info.func_src_ | exception | error | jax, pallas, closure, constants, tracing |
| The kernel function in the pallas_call {debug_info.func_src_ | exception | error | jax, pallas, kernel, return-value, validation |
| Cannot lower a pallas_call with constants. | exception | error | jax, pallas, lowering, constants, compiler |
| Only interpret mode is supported on CPU backend. | exception | error | jax, pallas, cpu, backend, interpret-mode |
| Mosaic GPU does not yet support AMD ROCm devices. Use ``comp | exception | error | jax, pallas, rocm, amd, gpu, backend, mosaic |
| The format string expects {n_placeholders} argument{'' if n_ | exception | error | pallas, format-string, debug-print, jax |
| run_scoped discharge does not support collective_axes yet. | exception | error | pallas, run-scoped, collectives, not-implemented, jax |
| Cannot handle new consts created by state discharge. | exception | error | pallas, state-discharge, run-scoped, not-implemented, jax |
| run_scoped lowering outside of Pallas does not support colle | exception | error | pallas, run-scoped, collectives, jax |
| get_global discharge is not supported in interpret mode. | exception | error | pallas, get-global, interpret-mode, not-implemented, jax |
| Cannot {name} on a non-semaphore Ref: {sem_aval} | exception | error | pallas, semaphore, type-validation, jax |
| Cannot {name} on a non-()-shaped semaphore: {sem_shape} | exception | error | pallas, semaphore, shape-validation, jax |
| Must {name} semaphores of the following types: {allowed_sema | exception | error | pallas, semaphore, dtype-validation, jax |
| Semaphore value shape {ref_value.shape} does not match aval | exception | error | pallas, semaphore, shape-mismatch, state-discharge, jax |
| Must signal an int32 value, but got {value_aval.dtype} | exception | error | pallas, semaphore, dtype-validation, jax |
| `device_id`s must be an int32 value, but got {aval.dtype} | exception | error | pallas, semaphore, dtype-validation, device-id, jax |
| Remote signal not implemented. | exception | error | pallas, semaphore, remote-signal, not-implemented, jax |
| Multiple core support not implemented. | exception | error | pallas, semaphore, multi-core, not-implemented, jax |
| Must wait an int32 value. | exception | error | pallas, semaphore, dtype-validation, jax |
| {axis_name} mixes JAX mesh and Pallas mesh grid axes | exception | error | pallas, mesh, collectives, axis-name, not-implemented, jax |
| `device_id_type` must be MESH if `device_id` is a dict, got: | exception | error | pallas, device-id, validation, jax |
| Number of device ids must match the number of mesh axes, but | exception | error | pallas, device-id, mesh, validation, jax |
| Unsupported device id type: {device_id_type} | exception | error | pallas, device-id, not-implemented, jax |
| TransformedRef must have been indexed before passing into ja | exception | error | pallas, refs, jaxpr-call, validation, jax |
| Only positional arguments are supported by debug_print on Pa | validation | error | jax, pallas, triton, debug-print, kwargs, gpu |
| unsupported dtypes: {x_aval.dtype} and {y_aval.dtype} | validation | error | jax, pallas, triton, jnp-minimum, dtype, notimplementederror |
| end must be greater than start, but got: {end} <= {start} | validation | error | jax, pallas, triton, arange, iota, empty-range, validation |
| start and end must fit in int32 | validation | error | jax, pallas, triton, int32-overflow, arange, large-tensors |
| cannot splat a tensor | validation | error | jax, pallas, triton, splat, scalar-shape, reshape, typeerror |
| cannot cast {src} tp {dst_type} | validation | error | jax, pallas, triton, dtype-cast, float8, notimplementederror |
| cannot cast from `{dtype_name}` | validation | error | jax, triton, pallas, dtype, cast, gpu |
| cannot cast to `{dtype_name}` | validation | error | jax, triton, pallas, dtype, cast, gpu |
| cannot cast {src} to {dst_type} | exception | error | jax, triton, pallas, mlir, cast, not-implemented |
| Only 2-argument concatenate is supported. | exception | error | jax, triton, pallas, concatenate, not-implemented |
| Only concatenate along the last dimension is supported. | exception | error | jax, triton, pallas, concatenate, axis, not-implemented |
| Only arguments with shape [..., 1] are supported. | exception | error | jax, triton, pallas, concatenate, shape, not-implemented |
| Only 2-argument stack is supported in Triton. | exception | error | jax, triton, pallas, stack, not-implemented |
| Only stack along the last dimension is supported in Triton. | exception | error | jax, triton, pallas, stack, axis, not-implemented |
| Only unstack of size 2 is supported in Triton. | exception | error | jax, triton, pallas, unstack, not-implemented |
| Only unstack along the last dimension is supported in Triton | exception | error | jax, triton, pallas, unstack, axis, not-implemented |
| Only power-of-2 num parts supported. | exception | error | jax, triton, pallas, split, not-implemented |
| Only equal-sized splits are supported. | exception | error | jax, triton, pallas, split, shape, not-implemented |
| Unexpected dim_block_size: {dim_block_size} | exception | error | jax, triton, pallas, internal, block-mapping, version-mismatch |
| unsupported cache modifier: {cache_modifier} | exception | error | jax, triton, pallas, load, cache-modifier, validation |
| unsupported eviction policy: {eviction_policy} | exception | error | jax, triton, pallas, load, eviction-policy, validation |
| loading from a block pointer is not supported | exception | error | jax, triton, pallas, load, block-pointer, not-implemented |
| unsupported pointer type: {ptr_type} | exception | error | jax, triton, pallas, load, pointer, validation |
| other requires mask to be provided | exception | error | jax, triton, pallas, load, mask, validation |
| other cannot be a block if pointer is not a block | exception | error | jax, triton, pallas, load, shape, validation |
| mask cannot be a block if pointer is not a block | exception | error | jax, triton, pallas, load, mask, shape, validation |
| Unsigned integer dtype {aval.dtype} is not supported for dot | exception | error | jax, pallas, triton, gpu, matmul, dtype |
| Unsupported dot algorithm: {precision}. | exception | error | jax, pallas, triton, precision, dot-algorithm |
| Unsupported dot precision: {precision}. | exception | error | jax, pallas, triton, precision, validation |
| a and b must be 2D, but got: {a_type.shape} and {b_type.shap | exception | error | jax, pallas, triton, shape, matmul |
| float64 dot requires M>=16, N>=8, K>=16 per warp tile (Trito | exception | error | jax, pallas, triton, float64, tile-shape |
| a and b must have the same element type, but got: {a_type.el | exception | error | jax, pallas, triton, dtype, matmul |
| Reductions with constants not supported. | exception | error | jax, pallas, triton, reduction, closures |
| `index_type` must be i32. | exception | error | jax, pallas, triton, argmax, index-dtype |
| `pallas` reduce operations only support one reduce axis. | exception | error | jax, pallas, triton, reduction, axes |
| Axis name {axis_name} not found in grid. | exception | error | jax, pallas, triton, grid, axis-index |
| Only single indexer is supported. | exception | error | jax, pallas, triton, atomics, indexing |
| unsupported atomic operation: {atomic_type} | exception | error | jax, pallas, triton, atomics |
| cmp and val must have identical dtypes and shapes | validation | error | jax, pallas, triton, dtype, atomic-cas |
| ref must be scalar. | validation | error | jax, pallas, triton, atomic-cas, shape |
| cmp must be scalar. | validation | error | jax, pallas, triton, atomic-cas, shape |
| val must be scalar. | validation | error | jax, pallas, triton, atomic-cas, shape |
| `next_power_of_2` requires a non-negative integer. | validation | error | jax, pallas, utils, validation |
| Scan with {num_extensive_outputs} extensive output(s) is not | validation | error | jax, pallas, triton, scan, control-flow |
| Scan with {num_extensive_inputs} extensive argument(s) is no | validation | error | jax, pallas, triton, scan, control-flow |
| erf_inv_lowering_helper not implemented for {x.dtype} | exception | error | jax, pallas, triton, dtype, erfinv |
| sign_lowering_helper not implemented for {x.dtype} | exception | error | jax, pallas, triton, dtype, sign |
| The two inputs to `nextafter` must have the same dtype, but | validation | error | jax, pallas, triton, dtype, nextafter |
| `nextafter` only supports float32 and float64, but got {x.dt | validation | error | jax, pallas, triton, dtype, nextafter |
| unreduced cannot contain None. All elements in unreduced sho | validation | error | jax, sharding, partition-spec, mesh |
| reduced cannot contain None. All elements in reduced should | validation | error | jax, sharding, partition-spec, mesh |
| `unreduced` and `reduced` argument to PartitionSpec cannot o | validation | error | jax, sharding, partition-spec, mesh |
| Expected unreduced_kind to be of type `jax.sharding.Unreduce | validation | error | jax, sharding, partition-spec, type-error |
| `unreduced_kind` should be `None` when `unreduced` is an emp | validation | error | jax, sharding, partition-spec, validation |
| partitions cannot overlap with unreduced axes passed to Part | validation | error | jax, sharding, partition-spec, mesh |
| partitions cannot overlap with reduced axes passed to Partit | validation | error | jax, sharding, partition-spec, mesh |
| Missing required keyword argument: 'in_sharding' | validation | error | jax, sharding, missing-argument, pjit |
| Context mesh cannot be empty. Please use `jax.set_mesh` API | validation | error | jax, mesh, sharding, context |
| layouts passed to `with_layout_constraint` must be of type ` | validation | error | jax, layout, type-validation |
| with_layout_constraint in eager mode can only be applied to | validation | error | jax, layout, eager-mode |
| All args passed to `explicit_layout` must have the same type | validation | error | jax, layout, consistency |
| Missing required keyword argument: 'in_layouts' | validation | error | jax, layout, missing-argument |
| 'devices' argument to pmap must be non-empty, or None. | validation | error | jax, pmap, devices, empty-argument |
| pmap in_axes must be an int, None, or (nested) container wit | validation | error | jax, pmap, in-axes, type-validation |
| pmap out_axes must be an int, None, or (nested) container wi | validation | error | jax, pmap, out-axes, type-validation |
| pmap requires at least one argument with a mapped axis. | validation | error | jax, pmap, in-axes, axis-size |
| pmapped function has static_broadcasted_argnums={static_broa | validation | error | jax, pmap, static-argnums, argument-count |
| pmap in_axes: value not a valid prefix (dynamic message incl | validation | error | jax, pmap, in-axes, pytree-prefix |
| Specified axis_size {axis_size} doesn't match received axis_ | validation | error | jax, pmap, axis-size, shape-mismatch |
| Only one profiler server can be active at a time. | validation | error | jax, profiler, singleton |
| No active profiler server. | validation | error | jax, profiler, lifecycle |
| Profile has already been started. Only one profile may be ru | error_code | error | jax, profiler, concurrency, singleton |
| Invalid trace folder: {latest_trace_folder} | validation | warning | jax, profiler, perfetto, filesystem |
| No profile started | error_code | error | jax, profiler, lifecycle |
| dirichlet requires alpha.ndim >= 1, got alpha.ndim == {} | validation | error | jax, random, dirichlet, input-validation, shape |
| dtype argument to `exponential` must be a float dtype, got { | validation | error | jax, random, exponential, dtype, input-validation |
| method argument to `gamma` must be one of {'exact', 'approxi | validation | error | jax, random, gamma, method, input-validation |
| dtype argument to `gamma` must be a float dtype, got {dtype} | validation | error | jax, random, gamma, dtype, input-validation |
| method argument to `loggamma` must be one of {'exact', 'appr | validation | error | jax, random, loggamma, method, input-validation |
| method argument to `poisson` must be one of {'exact', 'appro | validation | error | jax, random, poisson, method, input-validation |
| lam shape must be broadcastable to shape argument; got lam.s | validation | error | jax, random, poisson, broadcasting, shape |
| `poisson` with method='exact' is only implemented for the th | validation | error | jax, random, poisson, prng, not-implemented, tpu |
| dtype argument to `gumbel` must be a float dtype, got {dtype | validation | error | jax, random, gumbel, dtype, input-validation |
| Must provide valid mode for gumbel got: %s | validation | error | jax, random, gumbel, mode, input-validation |
| Number of samples without replacement ({k}) cannot exceed nu | validation | error | jax, random, categorical, sampling, top-k, shape |
| dtype argument to `laplace` must be a float dtype, got {dtyp | validation | error | jax, random, laplace, dtype, input-validation |
| dtype argument to `logistic` must be a float dtype, got {dty | validation | error | jax, random, logistic, dtype, input-validation |
| dtype argument to `pareto` must be a float dtype, got {dtype | validation | error | jax, random, pareto, dtype, input-validation |
| dtype argument to `t` must be a float dtype, got {dtype} | validation | error | jax, random, t-distribution, dtype, input-validation |
| method argument to `chisquare` must be one of {'exact', 'app | validation | error | jax, random, chisquare, method, input-validation |
| dtype argument to `chisquare` must be a float dtype, got {dt | validation | error | jax, random, chisquare, dtype, input-validation |
| dtype argument to `f` must be a float dtype, got {dtype} | validation | error | jax, random, f-distribution, dtype, input-validation |
| dtype argument to `maxwell` must be a float dtype, got {dtyp | validation | error | jax, random, maxwell, dtype, input-validation |
| dtype argument to `double_sided_maxwell` must be a float dty | validation | error | jax, random, dtype-validation |
| dtype argument to `weibull_min` must be a float dtype, got { | validation | error | jax, random, dtype-validation |
| dtype argument to `rayleigh` must be a float dtype, got {dty | validation | error | jax, random, dtype-validation |
| dtype argument to `wald` must be a float dtype, got {dtype} | validation | error | jax, random, dtype-validation |
| dtype argument to `geometric` must be an int dtype, got {dty | validation | error | jax, random, dtype-validation, discrete |
| dtype argument to `triangular` must be a float dtype, got {d | validation | error | jax, random, dtype-validation |
| dtype argument to `lognormal` must be a float or complex dty | validation | error | jax, random, dtype-validation |
| dtype argument to `binomial` must be a float dtype, got {dty | validation | error | jax, random, dtype-validation |
| {name} requires all arguments to have matching type. Got key | validation | error | jax, sharding, shard-map, pvary, random |
| PRNG key seed must be a scalar; got {seed!r}. | exception | error | jax, prng, seed, philox |
| PRNG key seed must be an integer; got {seed!r} | exception | error | jax, prng, seed, dtype-validation |
| philox2x32_random_bits got invalid prng key. | exception | error | jax, prng, philox, key-validation |
| requires 8-, 16-, 32- or 64-bit field width. | exception | error | jax, prng, philox, argument-validation |
| random bits array of size exceeding 2 ** 64 | exception | error | jax, prng, philox, shape-validation, resource-limits |
| PRNG key seed must be a scalar; got {seed!r}. | exception | error | jax, prng, seed, philox |
| PRNG key seed must be an integer; got {seed!r} | exception | error | jax, prng, seed, dtype-validation |
| philox4x32_random_bits got invalid prng key. | exception | error | jax, prng, philox, key-validation |
| requires 8-, 16-, 32- or 64-bit field width. | exception | error | jax, prng, philox, argument-validation |
| random bits array of size exceeding 2 ** 64 | exception | error | jax, prng, philox, shape-validation, resource-limits |
| PRNG with name {impl.name} already registered: {impl} | exception | error | jax, prng, registry, duplicate-registration |
| JAX encountered invalid PRNG key data: expected key_data to | exception | error | jax, prng, typeerror, input-validation |
| JAX encountered invalid PRNG key data: expected key_data.ndi | exception | error | jax, prng, ndim, typeerror |
| JAX encountered invalid PRNG key data: expected key_data.sha | exception | error | jax, prng, shape-mismatch, typeerror |
| JAX encountered invalid PRNG key data: expected key_data.dty | exception | error | jax, prng, dtype, uint32 |
| iteration over a 0-d key array | exception | error | jax, prng, iteration, unpacking |
| key array cannot be converted to boolean. | exception | error | jax, prng, truthiness, bool |
| JAX array with PRNGKey dtype cannot be converted to a NumPy | exception | error | jax, prng, numpy, serialization |
| random_unwrap takes key array operand, got {keys.dtype=} | exception | error | jax, prng, unwrap, dtype |
| _rbg_random_bits got invalid prng key. | exception | error | jax, prng, rbg, shape-mismatch |
| requires 8-, 16-, 32- or 64-bit field width. | exception | error | jax, prng, bit-width, dtype |
| Expected base_key to be a typed PRNG key; got {self._base_ke | exception | error | jax, prng, stateful-rng, typed-key |
| Expected counter to be a scalar integer ref; got {self._coun | exception | error | jax, prng, stateful-rng, ref, experimental |
| cannot operate on split stateful generator | exception | error | jax, prng, stateful-rng, split, spawn |
| When used within transformed code, jax.experimental.random.s | exception | error | jax, prng, stateful-rng, jit, tracing |
| PRNG key seed must be a scalar; got {seed!r}. | exception | error | jax, prng, seed, scalar, shape |
| PRNG key seed must be an integer; got {seed!r} | exception | error | jax, prng, seed, dtype, integer |
| threefry_2x32 requires uint32 arguments, got {} | exception | error | jax, prng, threefry, uint32, dtype |
| threefry_random_bits got invalid prng key. | exception | error | jax, prng, threefry, key-validation |
| requires 8-, 16-, 32- or 64-bit field width. | exception | error | jax, prng, threefry, bit-width, dtype |
| Only the type=1 case of eigh is implemented. | validation | error | jax, linalg, eigenvalue, not-implemented |
| Only the eigvals=None case of eigh is implemented. | validation | error | jax, linalg, eigenvalue, not-implemented |
| Expected 'output' to be either 'real' or 'complex', got {out | validation | error | jax, linalg, schur, argument-validation |
| Unsupported QR decomposition mode '{mode}' | validation | error | jax, linalg, qr-decomposition, argument-validation |
| mode must be 'right' or 'left', got {mode!r} | validation | error | jax, linalg, qr-decomposition, argument-validation |
| Array shapes are not compatible for Q @ c operation: a has s | validation | error | jax, linalg, shape-mismatch, qr-decomposition |
| Array shapes are not compatible for c @ Q operation: a has s | validation | error | jax, linalg, shape-mismatch, qr-decomposition |
| Expected assume_a to be one of {valid_assume_a}; got {assume | validation | error | jax, linalg, solve, argument-validation, version-migration |
| Invalid 'trans' value {trans} | validation | error | jax, linalg, solve-triangular, argument-validation |
| Expected A to be a (batched) square matrix, got {A.shape=}. | validation | error | jax, linalg, matrix-exponential, shape-mismatch |
| expected A to be a square matrix | validation | error | jax, linalg, internal-api, shape-mismatch |
| A.dtype={A.dtype} is not supported. | validation | error | jax, linalg, dtype, matrix-exponential |
| expected A to be a (batched) square matrix, got A.shape={A_a | validation | error | jax, linalg, frechet-derivative, shape-mismatch |
| expected E to be a (batched) square matrix, got E.shape={E_a | validation | error | jax, linalg, frechet-derivative, shape-mismatch |
| expected A and E to be the same shape, got A.shape={A_arr.sh | validation | error | jax, linalg, frechet-derivative, shape-mismatch |
| Arguments to jax.scipy.linalg.block_diag must have at most 2 | validation | error | jax, linalg, block-diagonal, shape-mismatch |
| diagonal and off-diagonal values must have same dtype, got { | validation | error | jax, linalg, tridiagonal, dtype |
| Only float32 and float64 inputs are supported as inputs to j | validation | error | jax, linalg, tridiagonal, dtype |
| for select='i', select_range must be specified. | validation | error | jax, linalg, tridiagonal, missing-argument |
| Got empty index range in select_range. | validation | error | jax, linalg, tridiagonal, argument-validation |
| eigh_tridiagonal(..., select='v') is not implemented | validation | error | jax, eigenvalues, not-implemented, linalg |
| 'select must have a value in {'a', 'i', 'v'}. | exception | error | jax, argument-validation, linalg, eigenvalues |
| method='qdwh' only supports mxn matrices where m < n where s | exception | error | jax, polar-decomposition, qdwh, shape-mismatch |
| Unknown polar decomposition method {method}. | exception | error | jax, argument-validation, polar-decomposition |
| The input `a` must be at least a 2-D array. | exception | error | jax, polar-decomposition, input-validation, shape-error |
| The argument `side` must be either 'right' or 'left'. | exception | error | jax, argument-validation, polar-decomposition |
| Blocked version is not implemented yet. | exception | warning | jax, sqrtm, not-implemented, linalg |
| Input 'T' must be square. | exception | error | jax, rsf2csf, schur-form, shape-validation |
| Input 'Z' must be square. | exception | error | jax, rsf2csf, schur-form, shape-validation |
| Input array shapes must match: Z: {Z_arr.shape} vs. T: {T_ar | exception | error | jax, rsf2csf, shape-mismatch |
| hankel: c must be at least 1-dimensional, got a scalar. | exception | error | jax, hankel, input-validation, shape-error |
| hankel: r must be at least 1-dimensional, got a scalar. | exception | error | jax, hankel, input-validation, shape-error |
| The length of f along the last axis must be at least 2; got | exception | error | jax, leslie-matrix, input-validation |
| Incorrect lengths for f and s. The length of s along the las | exception | error | jax, leslie-matrix, off-by-one, argument-length-validation |
| The length of `a` along the last axis must be at least 2; go | exception | error | jax, companion-matrix, polynomial, input-validation |
| fiedler_companion requires the last axis of 'a' to have nonz | exception | error | jax, fiedler-companion, polynomial, empty-array |
| n must be a positive integer; got {n}. | exception | error | jax, convolution-matrix, argument-validation |
| convolution_matrix: a must be at least 1-dimensional, got a | exception | error | jax, convolution-matrix, input-validation, shape-error |
| len(a) must be at least 1; got shape {a_arr.shape}. | exception | error | jax, convolution-matrix, empty-array |
| mode must be one of 'full', 'valid', 'same'; got {mode!r}. | exception | error | jax, convolution-matrix, argument-validation, invalid-enum-argument |
| Expected kind to be on of: {valid_kind}; got {kind} | exception | error | jax, scipy, linalg, argument-validation |
| Expected kind to be one of: {valid_kind}; got {kind} | exception | error | jax, scipy, linalg, argument-validation |
| n must be a positive power of 2; got {n}. | exception | error | jax, scipy, linalg, matrix, argument-validation |
| scale must be None, 'sqrtn', or 'n'; got {scale!r}. | exception | error | jax, scipy, linalg, dft, argument-validation |
| dtype must be a complex floating-point type; got {dtype}. | exception | error | jax, scipy, linalg, dtype, dft |
| Unrecognized method {method}. The two valid methods are eith | exception | error | jax, scipy, linalg, sylvester, argument-validation |
| Incompatible shapes for Sylvester equation: A: {A.shape} B: | exception | error | jax, scipy, linalg, shape-mismatch, sylvester |
| coordinates must be a sequence of length input.ndim, but {} | exception | error | jax, scipy, ndimage, coordinates, shape-mismatch |
| jax.scipy.ndimage.map_coordinates does not yet support mode | exception | error | jax, scipy, ndimage, unsupported-feature, boundary-mode |
| jax.scipy.ndimage.map_coordinates currently requires order<= | exception | error | jax, scipy, ndimage, interpolation, unsupported-feature |
| args argument to jax.scipy.optimize.minimize must be a tuple | exception | error | jax, scipy, optimize, type-error, argument-validation |
| Method {method} not recognized | exception | error | jax, scipy, optimize, argument-validation |
| in1 and in2 should have the same dimensionality | exception | error | jax, scipy, signal, fft, shape-mismatch |
| mode must be one of ['same', 'full', 'valid'] | exception | error | jax, scipy, signal, argument-validation |
| mapped axes must have same shape; got {in1.shape=} {in2.shap | exception | error | jax, scipy, signal, vmap, shape-mismatch |
| For 'valid' mode, One input must be at least as large as the | exception | error | jax, scipy, signal, convolution, shape-mismatch |
| Unrecognized {mode=} | exception | error | jax, scipy, signal, internal-api, argument-validation |
| mode must be one of ['full', 'same', 'valid'] | exception | error | jax, scipy, signal, convolution, argument-validation |
| in1 and in2 must have the same number of dimensions | exception | error | jax, scipy, signal, convolution, shape-mismatch |
| zero-size arrays not supported in convolutions, got shapes { | exception | error | jax, scipy, convolution, empty-array, shape-validation |
| One input must be smaller than the other in every dimension. | exception | error | jax, scipy, convolution, shape-mismatch |
| unsupported mode: {mode} | exception | error | jax, scipy, convolution, invalid-argument-value |
| Got {method=}; expected 'auto', 'fft', or 'direct'. | exception | error | jax, scipy, convolution, invalid-argument-value |
| convolve2d() only supports boundary='fill', fillvalue=0 | exception | error | jax, scipy, convolution, not-implemented, boundary |
| convolve2d() only supports 2-dimensional inputs. | exception | error | jax, scipy, convolution, rank-mismatch, shape-validation |
| correlate2d() only supports boundary='fill', fillvalue=0 | exception | error | jax, scipy, correlation, not-implemented, boundary |
| correlate2d() only supports 2-dimensional inputs. | exception | error | jax, scipy, correlation, rank-mismatch, shape-validation |
| overwrite_data argument not implemented. | exception | error | jax, scipy, detrend, not-implemented, immutable-arrays |
| Trend type must be 'linear' or 'constant'. | exception | error | jax, scipy, detrend, invalid-argument-value |
| Breakpoints must be non-negative and less than length of dat | exception | error | jax, scipy, detrend, breakpoints, index-out-of-range |
| The extension length n ({n}) is too big. It must not exceed | exception | error | jax, scipy, stft, signal-extension, shape-validation |
| Unknown value for mode {mode}, must be one of: ('psd', 'stft | exception | error | jax, scipy, internal-api, spectral-analysis |
| Unknown boundary option '{boundary}', must be one of: {list( | exception | error | jax, scipy, stft, boundary, invalid-argument-value |
| two-argument mode is available only when mode=='psd' | exception | error | jax, scipy, internal-api, spectral-analysis |
| two-arguments must have the same rank ({x.ndim} vs {y.ndim}) | exception | error | jax, scipy, csd, rank-mismatch, shape-validation |
| x and y cannot be broadcast together. | exception | error | jax, scipy, csd, broadcasting, shape-mismatch |
| nperseg must be a positive integer | exception | error | jax, scipy, stft, nperseg, jit-tracer, argument-validation |
| nfft must be greater than or equal to nperseg. | exception | error | jax, scipy, stft, fft-size, argument-validation |
| noverlap must be less than nperseg. | exception | error | jax, scipy, stft, overlap, argument-validation |
| betainc does not support complex-valued inputs. | validation | error | jax, scipy-special, complex-dtype, dtype-validation |
| dawsn does not support complex-valued inputs. | validation | error | jax, scipy-special, dawson, complex-dtype |
| entr does not support complex-valued inputs. | validation | error | jax, scipy-special, entropy, complex-dtype |
| kl_div does not support complex-valued inputs. | validation | error | jax, scipy-special, kl-divergence, complex-dtype |
| rel_entr does not support complex-valued inputs. | validation | error | jax, scipy-special, relative-entropy, complex-dtype |
| Riemann zeta function not implemented; pass q != None to com | exception | error | jax, scipy-special, zeta, not-implemented |
| Argument `n` to polygamma must be of integer type. Got dtype | validation | error | jax, scipy-special, polygamma, dtype-validation |
| polygamma does not support complex-valued inputs. | validation | error | jax, scipy-special, polygamma, complex-dtype |
| x.dtype={} is not supported, see docstring for supported typ | exception | error | jax, scipy-special, normal-cdf, dtype-validation |
| invalid value ({e.ty}) encountered in ndtri. | exception | error | jax, scipy-special, nan, ndtri, floating-point |
| series_order must be a Python integer. | exception | error | jax, scipy-special, log-ndtr, type-validation |
| series_order must be non-negative. | validation | error | jax, scipy-special, log-ndtr, argument-validation |
| series_order must be <= 30. | validation | error | jax, scipy-special, log-ndtr, argument-validation |
| x.dtype={np.dtype(dtype)} is not supported. | exception | error | jax, scipy-special, log-ndtr, dtype-validation |
| complex input not supported. | validation | error | jax, scipy-special, bessel, complex-dtype |
| Negative orders for normalization is not implemented yet. | exception | error | jax, scipy-special, legendre, not-implemented, spherical-harmonics |
| z.dtype={} is not supported, see docstring for supported typ | exception | error | jax, scipy-special, legendre, dtype-validation |
| z must be a 1D array. | validation | error | jax, scipy-special, legendre, not-implemented, argument-validation |
| Computations for m!=n are not yet supported. | exception | error | jax, scipy, legendre, not-implemented, argument-mismatch |
| The 'diff_n' argument to jax.scipy.special.sph_harm_y is not | exception | error | jax, scipy, spherical-harmonics, not-implemented, version-mismatch |
| expi does not support complex-valued inputs. | exception | error | jax, expi, complex-numbers, unsupported-dtype |
| Argument `x` to sici must be real-valued. Got dtype {x.dtype | exception | error | jax, sici, complex-numbers, unsupported-dtype |
| expn does not support complex-valued inputs. | exception | error | jax, expn, complex-numbers, unsupported-dtype, exponential-integral |
| exp1 does not support complex-valued inputs. | exception | error | jax, exp1, complex-numbers, unsupported-dtype, exponential-integral |
| x.dtype={dtype} is not supported, see docstring for supporte | exception | error | jax, spence, dtype, typeerror, dilogarithm |
| n must be a non-negative integer. | exception | error | jax, scipy, bernoulli, static-argument, input-validation |
| jnp.poch does not support complex-valued inputs. | exception | error | jax, scipy, poch, complex-dtype, unsupported-operation |
| hyp1f1 does not support complex-valued inputs. | exception | error | jax, scipy, hyp1f1, complex-dtype, unsupported-operation |
| Illegal nan_policy value {nan_policy!r}; expected one of {'p | exception | error | jax, scipy, stats, mode, nan-policy, argument-validation |
| Logic for `nan_policy` of {nan_policy} is not implemented | exception | error | jax, scipy, stats, mode, not-implemented, nan-policy |
| In order to best JIT compile `mode`, we cannot know whether | exception | error | jax, scipy, stats, mode, jit, not-implemented, nan-policy |
| In order to best JIT compile `rankdata`, we cannot know whet | exception | error | jax, scipy, stats, rankdata, jit, not-implemented, nan-policy |
| unknown method '{method}' | exception | error | jax, scipy, stats, rankdata, argument-validation, method |
| {nan_policy} is not supported | exception | error | jax, scipy, sem, nan-handling |
| `alpha` must be one-dimensional; got alpha.shape={alpha.shap | exception | error | jax, scipy, dirichlet, shape-validation |
| `x` must have either the same number of entries as `alpha` o | exception | error | jax, scipy, dirichlet, shape-validation |
| gaussian_kde does not support complex data | exception | error | jax, scipy, kde, complex-dtype |
| `dataset` input should have multiple elements. | exception | error | jax, scipy, kde, input-validation |
| `weights` input should be one-dimensional. | exception | error | jax, scipy, kde, weights, shape-validation |
| `weights` input should be of length n | exception | error | jax, scipy, kde, weights, length-mismatch |
| `bw_method` should be 'scott', 'silverman', a scalar, or a c | exception | error | jax, scipy, kde, bandwidth, argument-validation |
| mean does not have dimension {self.d} | exception | error | jax, scipy, kde, shape-validation |
| covariance does not have dimension {self.d} | exception | error | jax, scipy, kde, covariance, shape-validation |
| integrate_box_1d() only handles 1D pdfs | exception | error | jax, scipy, kde, dimensionality |
| the limits of integration in integrate_box_1d must be scalar | exception | error | jax, scipy, kde, scalar-validation |
| KDEs are not the same dimensionality | exception | error | jax, scipy, kde, dimensionality, shape-validation |
| only 1D box integrations are supported; use `integrate_box_1 | exception | error | jax, scipy, kde, not-implemented, api-parity |
| dynamically changing the bandwidth method is not supported | exception | error | jax, scipy, kde, not-implemented, api-parity |
| gaussian_kde does not support complex coordinates | exception | error | jax, scipy, kde, complex-dtype, not-implemented |
| points have dimension {}, dataset has dimension {} | exception | error | jax, scipy, kde, shape-validation, dimension-mismatch |
| points and xi must have same trailing dim | exception | error | jax, scipy, kde, shape-validation |
| precision matrix must match data dims | exception | error | jax, scipy, kde, precision-matrix, shape-validation |
| x and n must be of integer type; got x.dtype={x.dtype}, n.dt | exception | error | jax, scipy, multinomial, dtype, statistics |
| allow_singular argument of multivariate_normal.logpdf | exception | error | jax, scipy, multivariate-normal, not-implemented, covariance |
| multivariate_normal.logpdf got incompatible shapes | exception | error | jax, scipy, multivariate-normal, shape-mismatch |
| smap axis_name should be a `str` or a `Hashable`, but got {a | exception | error | jax, shard-map, smap, axis-name, distributed |
| smap in_axes must be an int, None, jax.sharding.Infer, or a | exception | error | jax, smap, shard-map, in-axes, distributed |
| smap in_axes must be an int, None, jax.sharding.Infer, or (n | exception | error | jax, smap, in-axes, pytree, validation |
| smap out_axes must be an int, None, or (nested) container wi | exception | error | jax, smap, out-axes, validation |
| shard_map requires a callable for its first argument, but go | exception | error | jax, shard-map, callable, type-error |
| in_specs passed to shard_map: {s} does not match the specs o | exception | error | jax, shard-map, sharding, partition-spec, distributed |
| shard_map out_specs rank error (msg from _spec_rank_error) | exception | error | jax, shard-map, out-specs, rank, distributed |
| shard_map out_specs vma error (msg from _inout_vma_error) | exception | error | jax, shard-map, out-specs, vma, replication |
| The context mesh cannot be empty. Use `jax.set_mesh(mesh)` t | exception | error | jax, shard-map, mesh, context, distributed |
| The context mesh {ctx_mesh} should match the mesh passed to | exception | error | jax, shard-map, mesh, context-mismatch, distributed |
| shard_map requires a `jax.sharding.Mesh` or a `jax.sharding. | exception | error | jax, shard-map, mesh, type-error, distributed |
| shard_map requires a non-empty mesh. Got {mesh} | exception | error | jax, shard-map, mesh, empty-mesh, distributed |
| `axis_names` argument of shard_map should be of type `frozen | exception | error | jax, shard-map, axis-names, type-error |
| jax.shard_map requires axis_names={axis_names} to be a subse | exception | error | jax, shard-map, axis-names, mesh, distributed |
| in_axes was not specified when axis_name={axis_names} was of | exception | error | jax, shard-map, smap, axis-types, in-specs |
| Invalid spec: {spec} | exception | error | jax, shard-map, partition-spec, manual-mode |
| unreduced/reduced can only be passed to {prefix}_specs when | exception | error | jax, shard-map, manual-mode, unreduced, experimental |
| unreduced in {prefix}_specs {s} can only be used when the me | exception | error | shard-map, partition-spec, mesh, jax, unreduced |
| reduced in {prefix}_specs {s} can only be used when the mesh | exception | error | shard-map, partition-spec, mesh, jax, reduced |
| shard_map in_specs argument must be a pytree of `jax.shardin | exception | error | shard-map, partition-spec, none-default, jax, typeerror |
| shard_map {prefix}_specs argument must refer to an axis mark | exception | error | shard-map, manual-axes, partition-spec, jax |
| shard_map {prefix}_specs argument must be a pytree of `jax.s | exception | error | shard-map, partition-spec, typeerror, jax |
| shard_map in_specs rank error (msg from _spec_rank_error) | exception | error | shard-map, rank-mismatch, partition-spec, jax |
| shard_map in_specs divisibility error (msg from _spec_divisi | exception | error | shard-map, divisibility, shape-mismatch, jax |
| in_specs containing unreduced {spec} passed to shard_map sho | exception | error | shard-map, unreduced, sharding-mismatch, jax |
| in_specs containing unreduced_kind {spec} passed to shard_ma | exception | error | shard-map, unreduced-kind, sharding-mismatch, jax |
| in_specs containing reduced {spec} passed to shard_map shoul | exception | error | shard-map, reduced, sharding-mismatch, jax |
| out_specs passed to shard_map should be equal to the unreduc | exception | error | shard-map, unreduced, out-specs, sharding-mismatch, jax |
| out_specs passed to shard_map should be equal to the unreduc | exception | error | shard-map, unreduced-kind, out-specs, sharding-mismatch, jax |
| out_specs passed to shard_map should be equal to the reduced | exception | error | shard-map, reduced, out-specs, sharding-mismatch, jax |
| Mesh shape of the input {a.sharding.mesh.shape_tuple} does n | exception | error | shard-map, mesh-mismatch, named-sharding, jax |
| Please pass `jax.Array`s with a `NamedSharding` as input to | exception | error | shard-map, abstract-mesh, named-sharding, jax |
| Invalid value ({e.ty}) encountered in sharded computation. | exception | error | shard-map, nan, inf, debug-flags, jax |
| Eager shard_map + unreduced/reduced + partial manual is not | exception | error | shard-map, eager, not-implemented, unreduced, jax |
| custom_vjp symbolic_zeros support with shard_map is not impl | exception | error | shard-map, custom-vjp, symbolic-zeros, autodiff, jax |
| device_put with explicit device not allowed within shard_map | validation | error | shard-map, device-put, placement, jax |
| Eager shard_map cannot return a `jax.Ref`. Please wrap your | validation | error | shard-map, ref, jit-required, jax |
| vmap spmd_axis_name cannot appear in shard_map in_specs | validation | error | jax, vmap, shard-map, spmd, partition-spec |
| vmapped away explicit mesh axis cannot appear in shard_map i | validation | error | jax, vmap, shard-map, explicit-mesh |
| vmap spmd_axis_name cannot appear in shard_map out_specs | validation | error | jax, vmap, shard-map, spmd, out-specs |
| vmapped away explicit mesh axis cannot appear in shard_map o | validation | error | jax, vmap, shard-map, explicit-mesh, out-specs |
| out_sharding passed to top_level_all_gather cannot be {out_s | validation | error | jax, shard-map, all-gather, sharding-type |
| Input sharding mesh {aval.sharding.mesh} should be equal to | validation | error | jax, mesh, all-gather, shard-map |
| multiple dimensions cannot be all_gathered since multi_dim=F | validation | error | jax, all-gather, shard-map, multi-dim |
| top_level_all_gather doesn't allow input {aval} to be unshar | validation | error | jax, all-gather, shard-map, resharding |
| top_level_all_gather maintains `top_level_all_gather(x, ...) | validation | error | jax, all-gather, shard-map, partition-spec |
| top_level_all_gather works when all mesh axes of context mes | validation | error | jax, all-gather, explicit-mesh, context-mesh |
| {self} implies that array axis {dim} is partitioned {p} time | validation | critical | jax, sharding, indivisible-shape, pjit, distributed |
| Subclasses should implement this method. | exception | error | jax, sharding, not-implemented, abstract-method |
| Subclasses should implement this method | exception | error | jax, sharding, not-implemented, memory-kind |
| Invalid dtype for `swap`. Ref dtype: {expected_out_ty.dtype} | validation | error | jax, dtype-mismatch, state-primitives, swap |
| `addupdate` must be called on `Ref` types: {ref_aval}. | validation | error | jax, state-primitives, type-error, addupdate |
| Invalid shape for `addupdate`. Ref shape: {ref_aval.shape}. | validation | error | jax, shape-mismatch, state-primitives, in-place-update |
| Invalid dtype for `addupdate`. Ref dtype: {ref_aval.dtype}. | validation | error | jax, dtype-mismatch, state-primitives, in-place-update |
| Invalid sharding for `addupdate`. Ref sharding: {ref_aval.sh | validation | error | jax, sharding, spmd, state-primitives |
| performing a set/swap operation with a differentiated value | exception | error | jax, autodiff, state-primitives, jvp |
| No support for vmapping over nontrivial slices just yet: {id | exception | error | jax, vmap, batching, not-implemented, slicing |
| Batching with multiple indexers not supported. | exception | error | jax, vmap, batching, not-implemented, multiple-indexing |
| Permutation {self.permutation} does not match the rank of th | validation | error | jax, transpose, shape-mismatch |
| Cannot transpose {x} to {self.permutation} | validation | error | jax, internal, type-dispatch |
| Cannot select {ref} | validation | error | jax, select, type-error |
| Cannot select from Refs of different types: {types} | validation | error | jax, multiref, type-mismatch |
| Cannot resolve attribute {name} from: {attrs} | exception | error | jax, multiref, attribute-mismatch |
| Multi-ref TransformedRef requires a single transform: {self} | exception | error | jax, multiref, api-restriction |
| {self!r} has no `shape`. | exception | error | jax, attribute-error, shape |
| {self!r} has no `dtype`. | exception | error | jax, attribute-error, dtype |
| Bitcast ref with dynamic size is not supported. | exception | error | jax, bitcast, dynamic-shape |
| Reshape ref with dynamic size is not supported. | exception | error | jax, reshape, dynamic-shape |
| Transpose with multiref is not supported. | exception | error | jax, transpose, multiref |
| Found inconsistent memory spaces in multiref: {self.ref} | exception | error | jax, memory-space, multiref |
| Aval memory space {inner_aval.memory_space} does not match t | exception | error | jax, memory-space, aval |
| len() of unsized object | exception | error | jax, len, unsized |
| {self!r} has no `sharding`. | exception | error | jax, sharding, attribute-error |
| {self!r} has no `manual_axis_type`. | exception | error | jax, axis-type, attribute-error |
| Bitcast 1D ref with bitwidth change is not supported. | exception | error | jax, bitcast, shape-restriction |
| Expected input and output shapes are the same after multiply | exception | error | jax, bitcast, divisibility |
| subset_by_index must be a tuple of size 2. | exception | error | jax, tpu, svd, linalg, argument-validation |
| Got empty index range in subset_by_index. | exception | error | jax, tpu, svd, linalg, off-by-one |
| Indices in subset_by_index must be non-negative. | exception | error | jax, tpu, svd, linalg, negative-index |
| Index in subset_by_index[1] exceeds matrix size. | exception | error | jax, tpu, svd, linalg, bounds-check |
| full_matrices and subset_by_index cannot be both be set. | exception | error | jax, tpu, svd, linalg, mutually-exclusive-args |
| The SVD algorithm parameter is not implemented on TPU. | exception | error | jax, tpu, svd, not-implemented, backend-support |
| Only the POLAR (which is also DEFAULT on TPU) SVD algorithm | exception | error | jax, tpu, svd, mlir-lowering, not-implemented |
| tpu_custom_call does not support non-trivial batching. | exception | error | jax, tpu, vmap, custom-call, not-implemented |
| invalid memory space: {self} | exception | error | jax, tpu, memory-space, enum, invalid-value |
| input_memory_space_colors only supports HBM, VMEM and SMEM | exception | error | jax, tpu, custom-call, serialization, not-implemented |
| invalid flag value: {value} | exception | error | tpu, pallas, custom-call, config-serialization, jax |
| Mosaic kernels cannot be automatically partitioned. Please w | exception | error | tpu, pallas, sharding, shard-map, jax |
| Replica lowering for Mosaic kernels not implemented. | exception | error | tpu, pallas, communication, replication, jax |
| Unknown core type: {core_type} | exception | error | tpu, pallas, core-type, version-mismatch, mlir |
| The iteration bounds and dimension semantics attributes must | exception | error | tpu, pallas, mlir, subkernel, validation |
| The iteration bound corresponding to the core-parallel dimen | exception | error | tpu, pallas, dynamic-shape, static-bounds, jax |
| A single Mosaic subkernel cannot contain multiple core shard | exception | error | tpu, pallas, core-parallel, subkernel, validation |
| The iteration bound corresponding to the core-parallel dimen | exception | error | tpu, pallas, core-parallel, subkernel, consistency |
| The manually assigned {collective_id=} in {kernel_name=} con | exception | error | tpu, pallas, collective-id, barrier, conflict |
| collective_id has to be specified when using a custom barrie | exception | error | tpu, pallas, collective-id, barrier, missing-argument |
| collective_id has to be unspecified or None when not using a | exception | error | tpu, pallas, collective-id, barrier, invalid-argument |
| vmem_limit_bytes must be an int: provided with a {type(vmem_ | exception | error | tpu, pallas, vmem, type-error, config |
| explicit tiling is only supported for SparseCore kernels. | exception | error | tpu, pallas, tiling, sparsecore, invalid-argument |
| explicit opt_level is only supported for SparseCore kernels. | exception | error | tpu, pallas, opt-level, sparsecore, invalid-argument |
| TPU generation is not supported | exception | error | tpu, hardware-generation, version-mismatch, jax |
| Unsupported TPU chip version: {chip_version} | exception | error | tpu, chip-version, unsupported-hardware, jax |
| Unsupported TPU device kind: {device_kind}. If you are not r | exception | error | tpu, cpu-testing, abstract-mesh, device-kind, jax |
| Lite chips, single core chips, and dual-core chips that do n | exception | error | tpu, chip-config, megacore, validation, jax |
| Backend {name} already initialized | exception | error | |
| Mock topology must be of the form "<number-of-slices> x <num | exception | error | |
| PJRT plugin config file should contain "library_path" field. | exception | error | |
| Multiple ':' separators for option in jax_pjrt_client_create | exception | error | |
| Cannot provide both 'factory' and 'options' when registering | exception | error | |
| Unknown backend: '{platform}' requested, but no platforms th | exception | error | |
| {err_msg} (set JAX_PLATFORMS='' to automatically choose an a | exception | error | |
| out_specs_fn already specified | validation | error | jax, colocated-python, api-misuse, configuration |
| out_specs already specified | validation | error | jax, colocated-python, api-misuse, configuration |
| devices already specified | validation | error | jax, colocated-python, devices, api-misuse |
| {op} has an unsupported layout: {out_layout_attr} | exception | error | jax, mosaic, gpu, layout, tiled-layout |
| Unsupported memory space: {orig_ref_ty.memory_space} | exception | error | jax, mosaic, gpu, memory-space, memref |
| Unsupported memory space: {ref_type.memory_space} | exception | error | jax, mosaic, gpu, memory-space, store |
| Expected TiledLayout, got {type(layout)} | exception | error | jax, mosaic, gpu, smem, layout |
| Transfer of {total_bits} bits is not divisible by {8 * utils | exception | error | jax, mosaic, gpu, smem, alignment, tma |
| `strides` must contain only 1s. | exception | error | jax, mosaic, gpu, vector, strides |
| Only slicing with static indices allowed. | exception | error | jax, mosaic, gpu, vector, static-index |
| Unsupported reduction kind: {op.kind} | exception | error | jax, mosaic, gpu, reduction, vector |
| Output layout {out_layout} must match the accumulator layout | exception | error | jax, mosaic, gpu, reduction, layout |
| Only 1 reduction dimension is supported. | exception | error | jax, mosaic, gpu, reduction |
| Unsupported layout: {src.layout} | exception | error | jax, mosaic, gpu, reduction, layout |
| Found multiple SwizzleTransformAttr | exception | error | jax, mosaic, gpu, swizzle, transforms |
| Unsupported transform: {transform} | exception | error | jax, mosaic, gpu, transforms, version-skew |
| Offsets {offsets} have lower rank than tiling {tiling} | exception | error | jax, mosaic, gpu, tiling, offsets, rank-mismatch |
| Offset {i} is not divisible by tile size {t} | exception | error | jax, mosaic, gpu, tiling, alignment |
| Strides {strides} have lower rank than tiling {tiling} | exception | error | jax, mosaic, gpu, tiling, strides, rank-mismatch |
| Can not tile strides when tiled dimensions have been transpo | exception | error | jax, mosaic, gpu, tiling, strides, transpose |
| Stride {s} is not divisible by {d} (tile size = {t}). Stride | exception | error | jax, mosaic, gpu, tiling, strides, alignment |
| SubViewOp only supports a single tile transform. | exception | error | jax, mosaic-gpu, memref, subview, tile-transform |
| Only support memref.cast where the input and output types ar | exception | error | jax, mosaic-gpu, memref, cast, type-mismatch |
| memref.cast transforms must have identical transforms for bo | exception | error | jax, mosaic-gpu, memref, cast, shared-memory, transforms |
| memref.cast tmem layouts must be identical for both input an | exception | error | jax, mosaic-gpu, tmem, cast, layout |
| Unsupported memory space when lowering memref.cast: {memory_ | exception | error | jax, mosaic-gpu, memref, cast, memory-space |
| Swizzle mismatch. In transforms swizzle: {in_swizzle}, out t | validation | error | jax, mosaic-gpu, transpose, swizzle, transforms |
| Size mismatch for in/out transforms. In transforms: {in_tran | validation | error | jax, mosaic-gpu, transpose, transforms, count-mismatch |
| Unsupported in/out transforms. In transform: {in_transform}, | exception | error | jax, mosaic-gpu, transpose, tile-transform, not-implemented |
| Cannot tile a transpose ({permutation}). Tiling dims ({permu | validation | error | jax, mosaic-gpu, transpose, tiling, permutation |
| Invalid in/out transforms. In transform: {in_transform}, out | validation | error | jax, mosaic-gpu, transpose, tiling, consistency |
| Expanding tiled dimensions is not supported. | exception | error | jax, mosaic-gpu, expand-shape, tiling, not-implemented |
| Expected the same number of in/out transforms, but got {in_t | validation | error | jax, mosaic-gpu, collapse-shape, transforms, count-mismatch |
| Only a single tiling transform is supported when collapsing | exception | error | jax, mosaic-gpu, collapse-shape, tile-transform, not-implemented |
| Collapsing the shape of a memref with non-contiguous strides | exception | error | jax, mosaic-gpu, collapse-shape, strides, contiguity |
| Input tiling {t_in.tiling} is not compatible with {op} | validation | error | jax, mosaic-gpu, collapse-shape, tiling, unsatisfiable |
| Input/output tiling mismatch when attempting to collapse a s | validation | error | jax, mosaic-gpu, collapse-shape, tiling, consistency |
| Reassociation {reassociation} is not compatible with tiling | validation | error | jax, mosaic-gpu, collapse-shape, reassociation, tiling |
| memref.LoadOp does not support transforms: {op} | exception | error | jax, mosaic-gpu, memref-load, transforms, not-implemented |
| memref.StoreOp does not support transforms: {op} | exception | error | jax, mosaic-gpu, memref-store, transforms, not-implemented |
| {ref} is not a memref. | validation | error | mosaic, gpu, tmem, mlir, type-validation |
| {ref} has a memory space {mem_ref_ty.memory_space} that is n | validation | error | mosaic, gpu, tmem, memory-space, mlir |
| {ref} has a layout {layout_attr} that does not match the exp | validation | error | mosaic, gpu, tmem, layout-mismatch |
| Expected both or neither of scales to be specified. | validation | error | mosaic, gpu, tcgen05, mma, scales, argument-validation |
| A custom return op must terminate the block. | validation | error | mosaic, gpu, custom-primitive, mlir, terminator |
| Unsupported op type: {op} | exception | error | mosaic, gpu, scf, control-flow, not-implemented |
| Input layouts {in_layouts} do not match yield layouts {yield | validation | error | mosaic, gpu, scf, while-loop, layout-mismatch |
| Output layouts {out_layouts} do not match condition layouts | validation | error | mosaic, gpu, scf, while-loop, layout-mismatch |
| Rule cannot handle an op with vector operands or results: {o | validation | error | mosaic, gpu, lowering, layout-mismatch, not-implemented |
| gpu.launch op not found. | validation | error | mosaic, gpu, gpu-launch, pass-pipeline, mlir |
| Tiles must have a decreasing rank | validation | error | mosaic, fragmented-array, tiling, layout, validation |
| Tiles must not be empty | validation | error | mosaic, fragmented-array, tiling, validation |
| Tile shape must only have positive sizes, got: {self.tiles} | validation | error | mosaic, fragmented-array, tiling, validation |
| Tiling {self.tiles} does not apply to shape {orig_shape} | validation | error | mosaic, fragmented-array, tiling, shape-mismatch |
| shape {orig_shape} is not a valid result of applying tiling | validation | error | mosaic, fragmented-array, tiling, shape-mismatch |
| Invalid dimension {dim} for tiling {self} | validation | error | mosaic, fragmented-array, tiling, index-out-of-range |
| Shape {shape} and strides {strides} must have the same lengt | exception | error | mosaic, fragmented-array, tiling, shape-strides, validation |
| `dimension` must be smaller than the rank of the array. | exception | error | jax, mosaic-gpu, iota, dimension-validation, shape-mismatch |
| Unsupported bitwidth: {bitwidth} | exception | error | jax, mosaic-gpu, layout-cast, bitwidth, wgmma, unsupported-dtype |
| Cannot convert from {self.layout} to {new_layout} | exception | error | jax, mosaic-gpu, layout-conversion, not-implemented, wgmma |
| Pointwise operations on {bitwidth}-bit types are unsupported | exception | error | jax, mosaic-gpu, pointwise, quantization, int4, int8, dtype-bitwidth |
| Cannot broadcast shape {self.shape} to layout {o.layout} | exception | error | jax, mosaic-gpu, broadcasting, splat-layout, shape-mismatch |
| Incompatible FragmentedArray layouts | exception | error | jax, mosaic-gpu, layout-mismatch, pointwise, fragmented-array |
| Incompatible FragmentedArray shapes | exception | error | jax, mosaic-gpu, shape-mismatch, registers, pointwise |
| trunc_div supports only integer types, got {self.mlir_dtype} | validation | error | jax, mosaic-gpu, division, dtype-check, integer-only |
| {arg_ty} | exception | error | mosaic, gpu, approx, internal |
| {vec_bitwidth} | exception | error | mosaic, gpu, ptx, packed-instruction |
| output_is_signed must be non-None if and only if the MLIR ty | validation | error | mosaic, gpu, bitcast, type-error, signedness |
| Only bitcast between types of the same bitwidth supported | validation | error | mosaic, gpu, bitcast, bitwidth |
| Only arrays with tiled layouts can be sliced | exception | error | mosaic, gpu, slicing, layout |
| Only slicing with static indices allowed | validation | error | mosaic, gpu, slicing, static-indices |
| Integer indexing not implemented for tiled dimensions (only | exception | error | mosaic, gpu, slicing, tiled-layout |
| Base indices of array slices must be aligned to the beginnin | validation | error | mosaic, gpu, slicing, tile-alignment |
| The slice shape must be a multiple of the tile shape. The ar | validation | error | mosaic, gpu, slicing, tile-alignment |
| Source and destination layouts aren't compatible for a broad | exception | error | jax, mosaic-gpu, layout, broadcast, fragmented-array |
| Splat layout does not support multimem | exception | error | jax, mosaic-gpu, splat-layout, multimem, tma |
| Atomic stores not supported for splat layout | exception | error | jax, mosaic-gpu, atomic, splat-layout, store |
| Only TiledLayouts support swizzling | exception | error | jax, mosaic-gpu, swizzle, strided-layout, store |
| Arrays with the splat layout can only be stored when they ha | exception | error | jax, mosaic-gpu, splat-layout, warpgroup, alignment, store |
| Multimem refs are not supported in store_tiled_async | exception | error | jax, mosaic, gpu, multimem, async-store, not-implemented |
| Replicated dimensions are not supported | exception | error | jax, mosaic, gpu, layout, replicated, warp, not-implemented |
| Unsupported register bitwidth: {reg_bitwidth} | exception | error | jax, mosaic, gpu, ptx, register-width, not-implemented |
| f32 not supported for async atomics | exception | error | jax, mosaic, gpu, atomics, f32, cluster-barrier, not-implemented |
| f32 only supports add atomics, got {atomic} | exception | error | jax, mosaic, gpu, atomics, f32, not-implemented |
| f16/bf16 not supported for async atomics | exception | error | jax, mosaic, gpu, atomics, f16, bf16, cluster-barrier, not-implemented |
| f16/bf16 only supports add, min, max atomics, got {atomic} | exception | error | jax, mosaic, gpu, atomics, f16, bf16, not-implemented |
| f16/bf16 SMEM/multimem atomics only support add, got {atomic | exception | error | jax, mosaic, gpu, atomics, f16, bf16, shared-memory, multimem, not-implemented |
| Unsupported element type for atomic stores: {element_type} | exception | error | jax, mosaic, gpu, atomics, dtype, not-implemented |
| Unsupported copy: {src.type} -> {dst.type} | exception | error | jax, mosaic-gpu, gpu-kernel, shared-memory, not-implemented |
| Need at least one array to concatenate | validation | error | jax, mosaic-gpu, validation, empty-input |
| {axis=} is out of bounds for array of {rank=} | validation | error | jax, mosaic-gpu, validation, axis-bounds |
| All arrays must have the same rank, got {len(arr.shape)} at | validation | error | jax, mosaic-gpu, validation, rank-mismatch |
| All arrays must have the same dtype, got {arr.mlir_dtype} at | validation | error | jax, mosaic-gpu, validation, dtype-mismatch |
| All arrays must have the same signedness, got {arr.is_signed | validation | error | jax, mosaic-gpu, validation, signedness |
| All arrays must have matching shapes along non-concatenated | validation | error | jax, mosaic-gpu, validation, shape-mismatch |
| All arrays must have the same layout, got {arr.layout} at in | validation | error | jax, mosaic-gpu, validation, layout-mismatch |
| Concatenating arrays with strided layout is only supported a | exception | error | jax, mosaic-gpu, not-implemented, wgmma, layout-limitation |
| Expected WGStridedFragLayout, got {arr.layout} at index {i} | validation | error | jax, mosaic-gpu, validation, layout-mismatch, wgmma |
| All WGStridedFragLayout arrays must have the same vec_size, | validation | error | mosaic-gpu, fragmented-array, layout-mismatch, concat, validation |
| Concatenating arrays with splat layout is not supported. | exception | error | mosaic-gpu, not-implemented, splat-layout, concat, fragmented-array |
| {op} does not have an {name} attribute | exception | error | mosaic-gpu, mlir, missing-attribute, layout-inference |
| {op} has {name} of an unexpected type: {result} | exception | error | mosaic-gpu, mlir, type-mismatch, attribute-type, layout-inference |
| Unknown attribute: {attr_name} | exception | error | mosaic-gpu, invalid-argument, attribute-dispatch, internal-api |
| Subclasses should override this method | exception | error | mosaic-gpu, not-implemented, abstract-method, memref-transform, subclassing |
| The collective size ({collective_size}) must divide the slic | validation | error | jax, mosaic-gpu, async-copy, cluster-partitioning, shape-validation |
| async_copy requires all GMEM strides except the last one to | validation | error | jax, mosaic-gpu, tma, memory-alignment, strides |
| Async copies only support striding up to 5 dimensions | validation | error | jax, mosaic-gpu, tma, rank-limit, shape-validation |
| None of the leading dimensions in the transformed slice shap | validation | error | jax, mosaic-gpu, tma, cluster-partitioning, shape-validation |
| Async copies require the number of bits copied along the las | validation | error | jax, mosaic-gpu, tma, memory-alignment, shape-validation |
| Async copies with {swizzle=} require the last dimension of t | validation | error | jax, mosaic-gpu, tma, swizzling, shared-memory |
| Expected same element type, got {element_type} and {dst_ref_ | validation | error | jax, mosaic-gpu, async-copy, dtype-mismatch |
| Only the TMA implementation supports reductions | validation | error | jax, mosaic-gpu, tma, reduction, unsupported-operation |
| Reduction op {reduction_op} not supported by the TMA impleme | validation | error | jax, mosaic-gpu, tma, reduction, dtype-mismatch, unsupported-operation |
| Barriers are required for TMA GMEM -> SMEM copies | validation | error | jax, mosaic-gpu, tma, barrier, synchronization |
| Only unit strides are supported but got {op.static_strides}. | validation | error | jax, mosaic-gpu, memref, layout-inference, strides, not-implemented |
| Transposed memrefs are not supported in ExpandShapeOp. | validation | error | jax, mosaic-gpu, memref, expand-shape, transpose, layout-inference |
| CollapseShapeOp with empty reassociation is not supported. | validation | error | jax, mosaic-gpu, memref, collapse-shape, layout-inference, not-implemented |
| CollapseShapeOp with non-contiguous strides is not supported | validation | error | jax, mosaic-gpu, memref, collapse-shape, strides, layout-inference |
| Only scalar memrefs are supported, got {ref_shape} | validation | error | jax, mosaic-gpu, memref, load-store, layout-inference, scalar |
| Unsupported transforms {transforms} | validation | error | jax, mosaic-gpu, smem-transforms, swizzle, tiling, layout-inference |
| Cannot apply {tiling_transform_str} with {swizzle_str} to me | validation | error | jax, mosaic-gpu, smem-transforms, swizzle, tiling, shape-mismatch |
| Unsupported TMA index shape {shape} | validation | error | jax, mosaic-gpu, tma, async-load, indices, gather |
| Only leading gather dimensions allowed. | validation | error | jax, mosaic-gpu, tma, gather, async-load, layout-inference |
| Only 2D gathers/scatters for async load/store are supported. | validation | error | jax, mosaic-gpu, tma, gather, async-load, rank, layout-inference |
| Cannot assign layout to async load with gather indices since | validation | error | jax, mosaic, gpu, layout, alignment, async-load |
| Expected the same number of in_{attr_suffix} ({len(in_layout | validation | error | jax, mosaic, mlir, layout, invariant |
| Expected the same number of out_{attr_suffix} ({len(out_layo | validation | error | jax, mosaic, mlir, layout, invariant |
| Inferred layout not found for operand {operand}. | exception | error | jax, mosaic, layout-inference, internal |
| Inferred layout {operand_layout} for operand {operand} does | exception | error | jax, mosaic, layout, conflict |
| No layout inference rule defined for {op} | exception | error | jax, mosaic, not-implemented, unsupported-op |
| Failed to infer a possible set of layouts. This should only | exception | error | jax, mosaic, layout, unsatisfiable-constraints |
| Unsupported layout for conversion from MLIR attribute: {attr | exception | error | jax, mosaic, layout, mlir-attribute, not-implemented |
| Unsupported transform {transform} | exception | error | jax, mosaic, transform, not-implemented |
| m_warps must be 1, 2, or 4, but got {m_warps=} | exception | error | jax, mosaic, mma, warps, validation |
| is_signed must be specified for integer types | exception | error | jax, mosaic, mma, dtype, signedness |
| M mismatch: {m} != {m2} | exception | error | jax, mosaic, mma, shape-mismatch |
| N mismatch: {n} != {n2} | exception | error | jax, mosaic, mma, shape-mismatch |
| K mismatch: {k} != {k2} | exception | error | jax, mosaic, mma, shape-mismatch |
| Dtype mismatch: {a.mlir_dtype} != {b.mlir_dtype} | exception | error | jax, mosaic, mma, dtype-mismatch |
| Unsupported operand type: {element_type} | exception | error | jax, mosaic, mma, unsupported-dtype |
| Only s32 accumulator supported for integer operands. | exception | error | jax, mosaic, mma, accumulator, int8 |
| Only signed accumulator supported for integer operands. | exception | error | jax, mosaic, mma, accumulator, signedness |
| {iterations=} must be positive | validation | error | profiling, mosaic, gpu, argument-validation |
| Unsupported trace scope: {trace_scope} | validation | error | profiling, mosaic, gpu, argument-validation |
| Scope {self.trace_scope} not supported | exception | error | profiling, mosaic, gpu, immutability |
| Block size is not a multiple of {scope_size} | validation | error | profiling, mosaic, gpu, cuda-launch-config |
| Allocated too many names | exception | error | profiling, mosaic, gpu, resource-exhaustion |
| Insufficient space to capture a full trace | exception | error | profiling, mosaic, gpu, buffer-overflow |
| Profiler collected an odd number of trace events. This likel | exception | error | profiling, mosaic, gpu, memory-corruption |
| Scope {spec.trace_scope} not supported | exception | error | profiling, mosaic, gpu, argument-validation |
| Scope {self.spec.trace_scope} not supported | exception | error | profiling, mosaic, gpu, immutability |
| Unsupported accumulator dtype: {acc_dtype} | exception | error | tcgen05, matmul, mosaic, gpu, dtype |
| Unsupported input dtype: {ty} | exception | error | tcgen05, matmul, mosaic, gpu, dtype |
| N must be a multiple of 8 and <= 256, got: {n} | validation | error | tcgen05, matmul, mosaic, gpu, shape-validation |
| M must be a multiple of 16 and <= 256, got: {m} | validation | error | tcgen05, matmul, mosaic, gpu, shape-validation |
| Unsupported scale type: {scale_type} | exception | error | tcgen05, mx-format, matmul, dtype |
| No swizzle is not supported | exception | error | tcgen05, matmul, swizzle, shared-memory |
| Either none or both scales should be provided | validation | error | tcgen05, matmul, mx-format, argument-validation |
| Sparse MMA not supported for M=64 | exception | error | jax, mosaic, gpu, sparse, mma, tensor-core, tcgen05, blackwell |
| Swizzle={b_swizzle} is too big for MMA with M=64. Try loweri | validation | error | jax, mosaic, gpu, mma, swizzle, tile-size, tcgen05 |
| Only M=128 and M=64 are supported for MMA, but got M={m} | validation | error | jax, mosaic, gpu, mma, shape-validation, tcgen05 |
| Sparse MMA unsupported for f32 | exception | error | jax, mosaic, gpu, sparse, dtype, mma, tcgen05 |
| MMA with element type {elem_type_str} does not support block | validation | error | jax, mosaic, gpu, block-scaling, dtype, mma, tcgen05, mxfp |
| MMA with element type {elem_type_str} only supports accumula | validation | error | jax, mosaic, gpu, accumulator-dtype, dtype, mma, tcgen05 |
| MMA with element type {elem_type_str} only supports accumula | validation | error | jax, mosaic, gpu, accumulator-dtype, dtype, mma, tcgen05 |
| Scale element type mismatch: expected f8e8m0fnu or f8e4m3fn, | validation | error | gpu, mosaic, tcgen05, dtype, block-scaling |
| Unsupported element type for block scaling: {a_element_type} | validation | error | gpu, mosaic, tcgen05, block-scaling, not-implemented |
| A scale shape mismatch: expected ({TMEM_ROWS}, {k_scales}), | validation | error | gpu, mosaic, tcgen05, shape-mismatch, block-scaling |
| A scale layout {a_scale.layout} is not supported | validation | error | gpu, mosaic, tcgen05, layout, tmem |
| Expected B scales to have a M=64 collective layout, got {b_s | validation | error | gpu, mosaic, tcgen05, layout, collective, cgmma |
| Expected B scales to have a M=128 layout, got {b_scale.layou | validation | error | gpu, mosaic, tcgen05, layout, block-scaling |
| B scale shape[0] must be a multiple of 128 and >= N={n * num | validation | error | gpu, mosaic, tcgen05, shape-mismatch, block-scaling, alignment |
| B scale shape mismatch: expected ({b_scale.shape[0]}, {k_sca | validation | error | gpu, mosaic, tcgen05, shape-mismatch, block-scaling |
| A sparse metadata shape mismatch: expected {(m, expected_met | validation | error | gpu, mosaic, tcgen05, sparsity, shape-mismatch, metadata |
| A sparse metadata dtype mismatch: expected i2, got {a_sparse | validation | error | gpu, mosaic, tcgen05, sparsity, dtype, metadata |
| 4-bit block scaled MMA only supports K-fastest operands, but | validation | error | gpu, mosaic, tcgen05, layout, mxfp4, block-scaling |
| 4-bit block scaled MMA only supports K-fastest operands, but | validation | error | gpu, mosaic, tcgen05, layout, mxfp4, block-scaling |
| B tiling too small. Increase swizzle or transpose the input. | validation | error | gpu, mosaic, tcgen05, sparsity, swizzle, not-implemented |
| A address calculation for multiple M tiles | validation | error | gpu, mosaic, tcgen05, tmem, not-implemented, mma |
| A sparse metadata address calculation for multiple tiles | validation | error | gpu, mosaic, tcgen05, sparsity, not-implemented, tiling |
| A scale address calculation for multiple M tiles | validation | error | gpu, mosaic, tcgen05, block-scaling, not-implemented, tiling |
| B scale address calculation for multiple N tiles | validation | error | gpu, mosaic, tcgen05, block-scaling, not-implemented, tiling |
| D address calculation for multiple M tiles | exception | error | gpu, mosaic, tcgen05, tmem, not-implemented, tiling |
| Minor dimension of shape must be divisible by packing, got: | exception | error | jax, mosaic, tmem, layout, shape-validation |
| Unsupported shape: {shape}. TMEM references must have either | exception | error | jax, mosaic, tmem, shape-validation, blackwell |
| Packing must be a power of 2, got: {packing} | exception | error | jax, mosaic, tmem, layout, argument-validation |
| Columns must be a multiple of 16, got: {columns} | exception | error | jax, mosaic, tmem, layout, alignment |
| tmem_addr_ref must be a memref or a pointer, got: {tmem_addr | exception | error | jax, mosaic, tmem, memref, argument-validation |
| tmem_addr_ref must be in shared memory, got: {addr_ref_ty} | exception | error | jax, mosaic, tmem, shared-memory, address-space |
| tmem_addr_ref must be an i32 memref, got: {addr_ref_ty} | exception | error | jax, mosaic, tmem, dtype, memref |
| tmem_addr_ref must contain a single element, got: {addr_ref_ | exception | error | jax, mosaic, tmem, memref, shape-validation |
| TMEM refs must have at least 32 rows, got: {shape[0]} | exception | error | jax, mosaic, tmem, shape-validation, hardware-constraint |
| collective argument must be provided when TMEM layout is inf | exception | error | jax, mosaic, tmem, missing-argument |
| TMEM can only be sliced, not indexed | exception | error | jax, mosaic, tmem, slicing, unsupported-operation |
| TMEM cannot be sliced along rows | exception | error | jax, mosaic, tmem, slicing, unsupported-operation |
| Cannot slice TMEM with multiple tiles along rows. | exception | error | jax, mosaic, tmem, slicing, layout, unsupported-operation |
| TMEM layout {self.layout} is not supported | exception | error | jax, mosaic, tmem, load, layout, unsupported-operation |
| Unsupported reduction for i32. Only min and max are supporte | exception | error | jax, mosaic, tmem, reduction, dtype, unsupported-operation |
| Unsupported reduction for f32. Only min, max, absmin, and ab | exception | error | mosaic, gpu, tcgen05, reduction, tensor-core, jax |
| Unsupported dtype for reduction: {self.dtype} | exception | error | mosaic, gpu, tcgen05, dtype, reduction, jax |
| Sparse meta layout loads unsupported. | exception | error | mosaic, gpu, tcgen05, sparse, tensor-memory, not-implemented, jax |
| Loading multiple row tiles | exception | error | mosaic, gpu, tcgen05, layout, tensor-memory, not-implemented, jax |
| Fused load-reduce is not supported for this layout | exception | error | mosaic, gpu, tcgen05, reduction, layout, jax |
| Loads from TMEM layout {self.layout} to register layout {lay | exception | error | mosaic, gpu, tcgen05, layout, tensor-memory, jax |
| TMEM stores expect a FragmentedArray, got: {value} | exception | error | mosaic, gpu, tcgen05, type-check, tensor-memory, jax |
| Stored array has shape {value.shape}, but TMEM has shape {se | exception | error | mosaic, gpu, tcgen05, shape-mismatch, tensor-memory, jax |
| Stored array has dtype {value.mlir_dtype}, but TMEM has dtyp | exception | error | mosaic, gpu, tcgen05, dtype-mismatch, tensor-memory, jax |
| Stored array has layout {value.layout}, but TMEM stores expe | exception | error | mosaic, gpu, tcgen05, layout, type-check, tensor-memory, jax |
| Sparse meta layout stores unsupported. | exception | error | mosaic, gpu, tcgen05, sparse, not-implemented, tensor-memory, jax |
| Storing from register layout {value.layout} to TMEM layout { | exception | error | mosaic, gpu, tcgen05, layout, tensor-memory, jax |
| Unsupported packing: {self.packing} | exception | error | mosaic, gpu, tcgen05, debug, packing, not-implemented, jax |
| Unsupported packing: {tmem_packing} for element type {elt_bi | exception | error | mosaic, gpu, tcgen05, packing, not-implemented, jax |
| Only {reg_packing} packing supported for bitwidth {elt_bitwi | exception | error | mosaic, gpu, tcgen05, packing, not-implemented, jax |
| Only {reg_packing} supported for element type {dtype}, but g | exception | error | mosaic, gpu, tcgen05, packing, not-implemented, jax |
| Unsupported reduction kind: {reduce} | exception | error | mosaic, gpu, tcgen05, reduction, internal, jax |
| K tile stride must be a multiple of 16 | validation | error | gpu, mosaic, tcgen05, sparse, shared-memory, alignment |
| Unsupported swizzle, expected 16, 32, 64 or 128, but got: {s | validation | error | gpu, mosaic, tcgen05, swizzle, validation |
| tcgen05.cp only supports fully packed TMEM references (packi | validation | error | gpu, mosaic, tcgen05, packing, tensor-memory |
| TMEM reference must have {TMEM_ROWS} rows, but got {tmem_ref | validation | error | gpu, mosaic, tcgen05, tensor-memory, shape-mismatch |
| Only standard TMEM layout is supported, got: {tmem_ref.layou | validation | error | gpu, mosaic, tcgen05, layout, tensor-memory |
| SMEM has shape {smem_shape}, but expected {expected_smem_sha | validation | error | gpu, mosaic, tcgen05, shape-mismatch, swizzle, shared-memory |
| The SMEM tiles must be contiguous | validation | error | gpu, mosaic, tcgen05, strides, shared-memory, contiguity |
| name must be non-empty | validation | warning | mosaic, gpu, io, dump, argument-validation |
| address_space not supported: {address_space} | exception | error | gpu, mosaic, address-space, nvptx, lowering |
| Pointer address space {ptr_ty.address_space} does not match | validation | error | gpu, mosaic, address-space, pointer, memref, llvm |
| Non-zero offset is not supported for ptr_as_memref | validation | error | gpu, mosaic, memref, offset, llvm |
| Empty array | validation | error | gpu, mosaic, tma, empty-array, argument-validation |
| Can't print the type {arg.type} | exception | error | gpu, mosaic, debug-print, unsupported-type, dtype |
| Cannot specify scope to a non-uniform debug_print. | validation | error | gpu, mosaic, debug-print, argument-validation |
| 2D+ vectors are not supported in debug_print: {vec_ty} | exception | error | gpu, mosaic, debug-print, vector, rank |
| Only 32-, 64- and 128-bit stores are supported | validation | error | gpu, mosaic, dsmem, multimem, store, bit-width |
| Folding {fold_rank} dimensions starting from {dim} is out of | validation | error | jax, mosaic-gpu, memref, shape-mismatch, index-out-of-bounds |
| strides={ref_ty.get_strides_and_offset()[0]}, {ref_ty.shape= | validation | error | jax, mosaic-gpu, memref, strides, layout |
| Can only infer one dimension | validation | error | jax, mosaic-gpu, memref, reshape, invalid-argument |
| Non-divisible unfold: | validation | error | jax, mosaic-gpu, memref, reshape, non-divisible-shape |
| Index {idx} along axis {axis} is out of bounds for shape {sh | validation | error | jax, mosaic-gpu, memref, index-out-of-bounds, indexerror |
| Strided slices not implemented | validation | error | jax, mosaic-gpu, memref, slice, not-implemented, strided |
| Slice {idx} along axis {axis} is out of bounds for shape {sh | validation | error | jax, mosaic-gpu, memref, slice, index-out-of-bounds |
| Expected an index-typed index | validation | error | jax, mosaic-gpu, memref, mlir, type-mismatch |
| Only up to 32 barriers per group supported | validation | error | jax, mosaic-gpu, barrier, hardware-limit, cuda |
| Barrier offset {offset} is out of bounds | validation | error | jax, mosaic-gpu, barrier, index-out-of-bounds |
| Expected a dynamic index or an integer, got {offset} | validation | error | jax, mosaic-gpu, barrier, type-mismatch |
| Unsupported scope: {scope} | validation | error | jax, mosaic-gpu, barrier, scope, invalid-argument |
| Can only await on CTA-local barriers | validation | error | jax, mosaic-gpu, barrier, scope, ptx |
| Only single-thread arrival is supported on pre-Hopper hardwa | validation | error | jax, mosaic-gpu, barrier, gpu-architecture, hopper |
| Predicate not supported for no-complete arrive | validation | error | jax, mosaic-gpu, barrier, predicate, not-implemented |
| arrive_expect_tx is only supported on Hopper+ hardware | validation | error | jax, mosaic-gpu, barrier, tma, gpu-architecture, hopper |
| complete_tx is only supported on Hopper+ hardware | validation | error | jax, mosaic-gpu, barrier, tma, gpu-architecture, hopper |
| Expected tiling to be at most rank of shape. Got tiling: {ti | validation | error | mosaic-gpu, tiling, shape-validation |
| Non-divisible tiling: | validation | error | mosaic-gpu, tiling, divisibility |
| Warp reduction group size should be a power of 2 (got {group | validation | error | mosaic-gpu, warp-reduction, power-of-two |
| {memref_ty} {static_offset=} is not divisible by {packing=}` | validation | error | mosaic-gpu, memref, alignment, sub-byte-types |
| Only 32-bit types supported | validation | error | mosaic-gpu, redux, dtype, hardware-limit |
| Only 32-bit scalar types supported | validation | error | mosaic-gpu, redux, dtype, hardware-limit |
| F32 redux only supported on Blackwell GPUs | validation | error | mosaic-gpu, redux, gpu-architecture, blackwell |
| Types must match, got {high.type} and {low.type} | validation | error | mosaic-gpu, prmt, type-mismatch |
| Can't bitcast {x.type} (of bitwidth {x_bw}) to {new_type} (o | validation | error | mosaic-gpu, bitcast, bitwidth |
| Can't bitcast {x.type} to {new_type} | validation | error | mosaic-gpu, bitcast, vector-types |
| Only 1D vectors are supported {v_ty} | validation | error | mosaic-gpu, vector-slice, rank, not-implemented |
| Cannot concatenate an empty list of vectors | validation | error | mosaic-gpu, vector-concat, empty-input |
| Cannot concatenate non-vector values | validation | error | mosaic-gpu, vector-concat, type-validation |
| Cannot concatenate vectors of different element types | validation | error | mosaic-gpu, vector-concat, dtype-mismatch |
| Can only concatenate 1D vectors | validation | error | mosaic-gpu, vector-concat, rank |
| Expected a memref type but got {ref} | validation | error | mosaic-gpu, memref, type-validation, shared-memory |
| Expected a single dimension when passing a single index | validation | error | jax, mosaic-gpu, cluster, api-misuse |
| Expected SMEM but got: {ref_ty.memory_space} | validation | error | jax, mosaic-gpu, smem, memory-space, gpu-clusters |
| Cannot retrieve the architecture without an insertion point | validation | error | jax, mosaic-gpu, mlir, insertion-point, architecture |
| Cannot retrieve the architecture: no module found | validation | error | jax, mosaic-gpu, architecture, mlir-module |
| WGMMA requires m and n to be multiples of 64 and 8, got {m} | validation | error | jax, mosaic-gpu, wgmma, gpu-kernel, shape-validation |
| PTX does not support unsigned WGMMA accumulators | validation | error | jax, mosaic-gpu, wgmma, ptx, signedness |
| Only WGMMA layouts supported in WGMMAAccumulator | validation | error | jax, mosaic-gpu, wgmma, fragmented-array, layout |
| Unsupported wgmma types {(out_ty, a_element_type)=} | validation | error | jax, mosaic-gpu, wgmma, dtype, tensor-core |
| Unsupported wgmma types {(out_ty, b_element_type)=} | validation | error | jax, mosaic-gpu, wgmma, dtype, tensor-core |
| Only f16 WGMMA supports transposes | validation | error | jax, mosaic-gpu, wgmma, transpose, dtype |
| Unsupported A register array dtype: {a.mlir_dtype} | validation | error | jax, mosaic-gpu, wgmma, dtype, registers |
| Unsupported A register array shape | validation | error | jax, mosaic-gpu, wgmma, shape, swizzle |
| Unsupported A register array layout | validation | error | jax, mosaic-gpu, wgmma, layout, fragmented-array |
| Unsupported WGMMA features with A in registers | validation | error | jax, mosaic-gpu, wgmma, api-misuse, registers |
| WGMMA instruction only supports f32, f16 and s32 out (got {o | validation | error | jax, mosaic-gpu, wgmma, dtype, accumulator |
| No swizzle is not supported | validation | error | jax, mosaic-gpu, wgmma, swizzle, smem |
| Existing output dtype must match preferred_element_type. | validation | error | jax, pallas, tpu, megablox, dtype-mismatch |
| group_offset must be a ()-shaped array. Got: {group_offset.s | validation | error | jax, pallas, tpu, megablox, shape-validation |
| No tuned tiling found for (m, k, n) = ({m}, {k}, {n}) | validation | error | jax, pallas, tpu, megablox, tiling, unsupported-shape |
| k_pages and v_pages must have the same shape. Got {k_pages.s | validation | error | jax, pallas, tpu, paged-attention, shape-validation |
| Number of Q heads must be divisible by number of KV heads. G | validation | error | jax, pallas, tpu, paged-attention, gqa, head-dim-validation |
| head_dim of Q must be the same as that of K/V. Got {head_dim | validation | error | jax, pallas, tpu, paged-attention, head-dim-validation |
| pages_per_compute_block must be divisible by pages per seque | validation | error | jax, pallas, tpu, paged-attention, layout-validation |
| `lengths` and `q` must have the same batch size | validation | error | jax, pallas, tpu, paged-attention, shape-validation |
| `page_indices` and `q` must have the same batch size | validation | error | jax, pallas, tpu, paged-attention, shape-validation |
| The dtype of `lengths` must be int32. Got {lengths.dtype} | validation | error | jax, pallas, tpu, paged-attention, dtype-validation |
| number of KV heads must be even when megacore_mode is 'kv_he | validation | error | jax, pallas, tpu, paged-attention, megacore, gqa |
| batch size must be even when megacore_mode is 'batch' | validation | error | jax, pallas, tpu, paged-attention, megacore, batch-size |
| megacore_mode must be one of ['kv_head', 'batch', None] | validation | error | jax, pallas, tpu, paged-attention, megacore, invalid-argument |
| {num_seqs[0]=} must be less or equal to {max_num_seqs=} | validation | error | jax, pallas, tpu, ragged-attention, capacity-validation |
| {pages_per_seq=} must be greater or equal to {min_pages_per_ | validation | error | jax, pallas, tpu, ragged-attention, kv-cache, capacity-validation |
| Total q tokens {cu_q_lens[num_seqs[0]]} must be less or equa | validation | error | jax, pallas, tpu, ragged-attention, cumulative-length-validation |
| {q_len=} must be less or equal to {kv_len=} at sequence {i}. | validation | error | jax, pallas, tpu, ragged-attention, length-validation |
| {num_seqs.shape=} must be (1,) | validation | error | jax, pallas, tpu, ragged-attention, shape-validation |
| Q head_dim {head_dim} must be the same as that of K/V {head_ | validation | error | jax, pallas, tpu, ragged-attention, head-dim-validation |
| Expected {kv_lens.shape=} to be ({max_num_seqs},) where `max | validation | error | jax, pallas, tpu, paged-attention, shape-validation |
| Expected {cu_q_lens.shape=} to be ({max_num_seqs + 1},) whe | validation | error | jax, pallas, tpu, ragged-attention, shape-validation |
| The dtype of `kv_lens`, `page_indices`, and `cu_q_lens` must | validation | error | jax, pallas, tpu, dtype-validation, paged-attention |
| {num_q_heads=} must be divisible by {num_kv_heads=} | validation | error | jax, pallas, tpu, gqa, attention, head-dimension |
| {sliding_window=} must be positive. | validation | error | jax, pallas, tpu, sliding-window, config-validation |
| {soft_cap=} must not be 0.0. | validation | error | jax, pallas, tpu, softcap, attention, config-validation |
| {num_kv_pages_per_block=} must be in range (0, {pages_per_se | validation | error | jax, pallas, tpu, block-size, paged-attention |
| {num_queries_per_block=} must be positive. | validation | error | jax, pallas, tpu, block-size, validation |
| {vmem_limit_bytes=} must be positive. | validation | error | jax, pallas, tpu, memory-limit, validation |
| Not implemented: {num_combined_kv_heads=} can not be XLA ful | validation | error | jax, pallas, tpu, attention, head-tiling, not-implemented |
| TPU version must be 4 or higher. | exception | error | jax, pallas, tpu, hardware-version, not-implemented |
| Shape too large: {np.prod(shape)} > {np.iinfo(jnp.uint32).ma | validation | error | jax, pallas, tpu, random, philox, shape-limit |
| Shape dimension {shape[-2:]} must be divisible by {block_siz | validation | error | jax, pallas, tpu, random, philox, alignment |
| Offset must be scalar, got {offset.shape} | validation | error | jax, pallas, tpu, random, philox, scalar-validation |
| Only 32-bit PRNG supported. | validation | error | jax, pallas, tpu, random, philox, bit-width |
| Shape too large: {np.prod(shape)} > {np.iinfo(jnp.uint32).ma | validation | error | jax, pallas, tpu, random, threefry, shape-limit |
| Shape dimension {shape[-2:]} must be divisible by {block_siz | validation | error | jax, pallas, tpu, random, threefry, alignment |
| Only 32-bit PRNG supported. | validation | error | jax, pallas, tpu, random, threefry, bit-width |
| Higher-order AD not supported. | exception | error | jax, pallas, tpu, splash-attention, autodiff, higher-order-ad |
| Block sizes for dq kernel are not needed with a fused kernel | validation | error | jax, pallas, tpu, splash-attention, config-validation |
| Mask function must return a boolean-valued array, but got: { | validation | error | jax, pallas, tpu, splash-attention, dtype, mask |
| block_kv must be a multiple of {NUM_LANES} | exception | error | jax, pallas, tpu, splash-attention, segment-ids, block-size |
| block_q must be a multiple of {NUM_LANES} | exception | error | jax, pallas, tpu, splash-attention, segment-ids, block-size |
| {bkv_compute=} should be a multiple of {NUM_LANES} | exception | error | jax, pallas, tpu, splash-attention, block-size |
| partial_mask_blocks must be of type np.bool_ but got {partia | validation | error | jax, pallas, tpu, splash-attention, dtype, mask |
| Expected {expected_kv_rank}-dim 'key' tensor for MQA. Instea | validation | error | jax, pallas, tpu, splash-attention, shape, rank |
| Expected 'key' head dimension to be: {head_dim_qk}. Instead | validation | error | jax, pallas, tpu, splash-attention, shape, head-dim |
| In MHA, expected number of 'key' heads ({num_kv_heads}) to b | validation | error | jax, pallas, tpu, splash-attention, gqa, shape |
| Expected 'key' {k.shape} and 'value' {v.shape} to have the s | validation | error | jax, pallas, tpu, splash-attention, shape |
| {bkv=} must be a multiple of {bkv_compute=}. | validation | error | jax, pallas, tpu, splash-attention, block-size |
| {bkv_compute=} must be a multiple of {NUM_LANES}. | validation | error | jax, pallas, tpu, splash-attention, block-size |
| Invalid shape for q segment_ids: {segment_ids.q.shape}. Expe | validation | error | jax, pallas, tpu, splash-attention, segment-ids, shape |
| Invalid shape for kv segment_ids: {segment_ids.kv.shape}. Ex | validation | error | jax, pallas, tpu, splash-attention, segment-ids, shape |
| Higher-order AD not supported | exception | error | jax, pallas, tpu, splash-attention, autodiff |
| {bq=} should not be greater than {q_seq_len=} | validation | error | jax, pallas, tpu, splash-attention, block-size, sequence-length |
| {bkv=} should not be greater than {kv_seq_len=} | validation | error | jax, pallas, tpu, splash-attention, block-size, sequence-length |
| {bkv=} must be a multiple of {NUM_LANES}. | validation | error | jax, pallas, tpu, splash-attention, block-size, backward |
| {bkv_compute=} should not be greater than {bkv=} | validation | error | jax, tpu, pallas, splash-attention, block-sizes, validation |
| {bkv=} should be a multiple of {bkv_compute=} | validation | error | jax, tpu, pallas, splash-attention, block-sizes, divisibility |
| Need to specify backward blocks. | validation | error | jax, pallas, splash-attention, block-sizes, backward |
| The sharding must divide the mask blocks evenly between devi | validation | error | jax, tpu, splash-attention, sharding, dynamic-mask |
| Sharding the kv sequence dimension is not supported | validation | error | jax, tpu, splash-attention, sharding, partition-spec |
| Unexpected mask shape: {mask.shape} | validation | error | jax, splash-attention, mask, shape-validation |
| Invalid shape for other: {other.shape}, expected: {self.shap | exception | error | jax, splash-attention, mask, shape-mismatch |
| chunk_size must be positive | validation | error | jax, splash-attention, mask, chunked-attention, argument-validation |
| Masks must have the same shape | validation | error | jax, splash-attention, mask, shape-mismatch |
| Unsupported empty tuple of masks | validation | error | jax, splash-attention, mask, empty-collection, validation |
| Unexpected mask shape, got: {mask.shape}, expected: {shape} | validation | error | jax, splash-attention, mask, shape-mismatch, multi-head |
| masks should be of type Mask | validation | error | jax, splash-attention, mask, type-validation |
| Nesting MultiHeadMasks is not supported | validation | error | jax, splash-attention, mask, multi-head, nesting |
| {q_block_size=} should divide {q_seq_len_per_shard=}. | validation | error | jax, tpu, splash-attention, shape-mismatch, pallas |
| {head_shards=} should divide {head_count=}. | validation | error | jax, tpu, splash-attention, sharding, shape-mismatch |
| Expected int32 input, but got {array.dtype}. | validation | error | jax, dtype, splash-attention, tpu, int64 |
| Expected a 3-dim mask, instead got: {mask.shape=} | validation | error | jax, splash-attention, mask, rank-mismatch, tpu |
| bfloat16 support not implemented for LSTM | exception | error | jax, lstm, rnn, bfloat16, cudnn, gpu |
| Unexpected precision specifier value {precision} | exception | error | jax, lstm, rnn, invalid-argument, precision |
| `seq_lengths` can only be int32. | exception | error | jax, lstm, rnn, dtype, int32 |
| Dropout not supported in LSTM reference because we cannot de | exception | error | jax, lstm, rnn, dropout, not-implemented |
| Unsupported shape: {shape} | exception | error | jax, sparse, coo, gpu, shape-validation |
| todense_transpose for {type(obj)} | exception | error | jax, sparse, autodiff, todense, csr |
| sparse_format={sparse_format!r} not recognized; must be one | exception | error | jax, sparse, argument-validation, format-string |
| batch_dims must be None or satisfy 0 < dim < n_batch. Got {b | exception | error | jax, sparse, bcoo, vmap, batching |
| data batch dimensions not compatible for {data.shape=}, {sha | exception | error | jax, sparse, bcoo, shape-validation, constructor |
| Invalid {data.shape=} for {nse=}, {n_batch=}, {n_dense=} | exception | error | jax, sparse, bcoo, shape-validation, nse |
| indices batch dimensions not compatible for {indices.shape=} | exception | error | jax, sparse, bcoo, indices, shape-validation |
| Invalid ={indices.shape=} for {nse=}, {n_batch=}, {n_dense=} | exception | error | jax, sparse, bcoo, indices, invariant |
| Expected 0 < bdim <= n_batch; got {bdim=}, {n_batch=} | exception | error | jax, sparse, bcoo, vmap, fromdense |
| First argument to bcoo_extract should be a BCOO array. Got { | exception | error | jax, sparse, type-check, bcoo |
| shape mismatch: {sparr.shape=} {a.shape=} | exception | error | jax, sparse, shape-validation, bcoo |
| transpose of bcoo_extract with assume_unique=False | exception | error | jax, sparse, bcoo, autodiff, transpose |
| bcoo_extract with unequal batch dimensions. | exception | error | jax, sparse, bcoo, vmap, batching |
| {batch_dims=} out of range for indices with {n_batch=} | exception | error | jax, sparse, bcoo, vmap, batching |
| transpose permutation must be a tuple/list/ndarray, got {typ | exception | error | jax, sparse, bcoo, transpose, type-check |
| transpose permutation isn't a permutation of operand dimensi | exception | error | jax, sparse, bcoo, transpose, permutation |
| transpose permutation cannot permute batch axes with non-bat | exception | error | jax, sparse, bcoo, transpose, permutation, batching |
| bcoo_slice: input should be BCOO array, got type(mat)={type( | exception | error | jax, sparse, bcoo, type-error, slice |
| bcoo_slice: indices must have size mat.ndim={mat.ndim} | exception | error | jax, sparse, bcoo, validation, shape-mismatch |
| len(strides) = {len(strides)}; expected {mat.ndim} | exception | error | jax, sparse, bcoo, strides, validation |
| strides must be a sequence of positive integers; got {stride | exception | error | jax, sparse, bcoo, strides, positive-value-required |
| bcoo_slice: invalid indices. Got {start_indices=}, {limit_in | exception | error | jax, sparse, bcoo, index-out-of-bounds, slice |
| bcoo_dynamic_slice: indices must have size mat.ndim={mat.ndi | exception | error | jax, sparse, bcoo, validation, shape-mismatch |
| slice_sizes must be less than or equal to operand shape, got | exception | error | jax, sparse, bcoo, window-too-large, dynamic-slice |
| bcoo_multiply_sparse: arrays must have same number of dimens | exception | error | jax, sparse, bcoo, shape-mismatch, elementwise-multiply |
| bcoo_multiply_sparse: arrays with differing numbers of dense | exception | error | jax, sparse, bcoo, not-implemented, dense-dimensions |
| BCSR sparse.empty: must have 2 sparse dimensions. | exception | error | jax, sparse, bcsr, shape-validation |
| BSCR.from_bcoo requires n_sparse=2; got {arr.n_sparse=} | exception | error | jax, sparse, bcsr, bcoo, format-conversion |
| BCSR from_scipy_sparse with nonzero n_dense/n_batch. | exception | error | jax, sparse, bcsr, scipy, not-implemented |
| BCSR from_scipy_sparse requires 2D array; {mat.ndim}D is giv | exception | error | jax, sparse, bcsr, scipy, input-validation |
| Cannot map in_axis={axis} for BCSR array with n_batch={val.n | exception | error | jax, sparse, bcsr, vmap, batching |
| BCSR: cannot add out_axis={axis} for BCSR array with n_batch | exception | error | jax, sparse, bcsr, vmap, out-axes |
| COO must have ndim=2; got {shape=} | exception | error | jax, sparse, coo, shape-validation |
| axes argument to transpose() | exception | error | jax, sparse, coo, transpose, not-implemented |
| COO.tree_unflatten: invalid {aux_data=} | exception | error | jax, sparse, coo, pytree, serialization |
| matmul between two sparse objects. | exception | error | jax, sparse, coo, matmul, not-implemented |
| matmul with object of shape {other.shape} | exception | error | jax, sparse, coo, matmul, shape-validation |
| CSR must have ndim=2; got {shape=} | exception | error | jax, sparse, csr, shape-validation |
| matmul between two sparse objects. | exception | error | jax, sparse, csr, matmul, not-implemented |
| matmul with object of shape {other.shape} | exception | error | jax, sparse, csr, matmul, shape-validation |
| CSR.tree_unflatten: invalid {aux_data=} | exception | error | jax, sparse, csr, pytree, serialization |
| CSC must have ndim=2; got {shape=} | exception | error | jax, sparse, csc, shape-validation |
| CSC.tree_unflatten: invalid {aux_data=} | exception | error | jax, sparse, pytree, csc, serialization |
| Cannot transpose with respect to sparse indices | exception | error | jax, sparse, csr, autodiff, gradient |
| must have search dim > 0, got {k} | exception | error | jax, sparse, lobpcg, eigensolver, validation |
| expected search dim * 5 < matrix dim (got {k * 5}, {n}) | exception | error | jax, sparse, lobpcg, validation |
| A, X must have same dtypes (were {test_output.dtype}, {dt}) | exception | error | jax, sparse, lobpcg, dtype |
| A must be ({n}, {n}) matrix A, got output {s} | exception | error | jax, sparse, lobpcg, shape-mismatch |
| data types do not match: {data.dtype=} {b.dtype=} | exception | error | jax, sparse, spsolve, dtype |
| index arrays must be integer typed; got {indices.dtype=} {in | validation | error | jax, sparse, spsolve, dtype, indices |
| Arrays must be one-dimensional. Got {data.shape=} {indices.s | validation | error | jax, sparse, spsolve, shape, dimensions |
| Invalid CSR buffer sizes: {data.shape=} {indices.shape=} {in | validation | error | jax, sparse, spsolve, csr, buffer-sizes |
| {reorder=} not valid, must be one of [1, 2, 3, 4] | validation | error | jax, sparse, spsolve, gpu, parameter-validation |
| Invalid {n_batch=}, {n_dense=} for {shape=} | validation | error | jax, sparse, bcoo, random, validation |
| got {nse=}, expected to be between 0 and {sparse_size} | validation | error | jax, sparse, bcoo, random, nse |
| {indices_dtype=} does not have enough range to generate spar | validation | error | jax, sparse, bcoo, random, index-overflow |
| sparse rule for {primitive} is not implemented because it wo | exception | error | jax, sparse, sparsify, lax, not-implemented |
| sparse rule for {primitive} is not implemented. | exception | error | jax, sparse, sparsify, not-implemented, lax |
| sparse rule for {lax.integer_pow_p} with non-positive expone | exception | error | jax, sparse, sparsify, power, densifying |
| Addition between sparse matrices of different shapes. | exception | error | jax, sparse, sparsify, addition, shape-mismatch |
| Addition between sparse matrices with different batch/dense | exception | error | jax, sparse, sparsify, addition, batch-dims |
| Error reading persistent compilation cache entry for '{cache | console | warning | jax, compilation-cache, persistence, io |
| Error reading persistent compilation cache entry for '{modul | console | warning | jax, compilation-cache, deserialization, io |
| Error writing persistent compilation cache entry for '{modul | console | warning | jax, compilation-cache, io, permissions |
| TRACER_LEAK_DEBUGGER_WARNING | console | warning | jax, tracer-leak, debugging, memory |
| JAX_SLICE_INDEX has been deprecated. Please use JAX_PARTITIO | console | warning | jax, distributed, environment-variable, deprecation |
| `slice_index` has been deprecated. Please use `partition_ind | console | warning | jax, distributed, api-rename, deprecation |
| Explicitly requested dtype {}{} is not available, and will b | console | warning | jax, dtype, precision, x64 |
| When at least one mesh axis of `pred` is in auto mode, calli | console | warning | jax, distributed, spmd, sharding, error-check |
| cuDNN not found. | exception | error | jax, cuda, cudnn, gpu, shared-library |
| Argument to register_custom_call_partitioner was not a pjrt_ | validation | error | jax, xla, custom-call, capsule, validation |
| This operation is implemented for a PjRt-compatible backend | exception | error | jax, dlpack, ifrt, pjrt, interop |
| DLPack is supported for PjRt-compatible backends only. | exception | error | jax, dlpack, ifrt, pjrt, interop |
| DLPack is only supported for devices addressable by the curr | exception | error | jax, dlpack, device, multi-process |
| dtype parameter is not supported by Buffer.__array__. | validation | error | jax, ffi, numpy, dtype |
| Buffer.__array__ with copy=True is not supported. | validation | error | jax, ffi, numpy, copy |
| Buffer.__dlpack__ with copy=True is not supported. | validation | error | jax, ffi, dlpack, copy |
| Unsupported index dtype: %s | validation | error | jax, cuda, cusparse, sparse, dtype |
| Unsupported data dtype: %s | validation | error | jax, gpu, sparse, dtype |
| Init is not implemented | exception | error | jax, collectives, cpu, api-misuse |
| Finalize is not implemented | exception | error | jax, collectives, cpu, api-misuse |
| make_gloo_tcp_collectives only implemented for linux and mac | exception | error | jax, gloo, collectives, platform-support |
| make_mpi_collectives is not implemented for Windows | exception | error | jax, mpi, collectives, platform-support |
| c_api argument to load_pjrt_plugin is not a pjrt_c_api capsu | validation | error | jax, pjrt, plugin, capsule, validation |
| Argument to get_c_api_topology was not a pjrt_c_api capsule. | validation | error | jax, pjrt, topology, capsule, validation |
| Argument to get_c_api_topology contained a null pointer. | validation | error | jax, pjrt, topology, capsule, null-pointer |
| get_topology_for_devices requires >= 1 devices. | validation | error | jax, topology, devices, validation |
| Expected tuple, got %s. | validation | error | jax, pytree, type-mismatch, tuple |
| Tuple arity mismatch: %d != %d; tuple: %s. | validation | error | jax, pytree, arity-mismatch, tuple |
| Expected list, got %s. | validation | error | jax, pytree, type-mismatch, list |
| List arity mismatch: %d != %d; list: %s. | validation | error | jax, pytree, arity-mismatch, list |
| Expected dict, got %s. | validation | error | jax, pytree, type-mismatch, dict |
| Dict key mismatch; expected keys: %s; present keys: %s. | validation | error | jax, pytree, dict, key-mismatch |
| Expected named tuple, got %s. | validation | error | jax, pytree, type-mismatch, namedtuple |
| Malformed pytree proto (invalid node type) | validation | error | jax, pytree, serialization, protobuf |
| Could not find type: %s. | exception | error | jax, pytree, pickle, registry |
| Pytree serialization too large to deserialize. | exception | error | jax, pytree, protobuf, size-limit |
| Could not deserialize PyTreeDefProto. | exception | error | jax, pytree, protobuf, corrupt-data |
| Malformed pickled PyTreeDef, expected 2-tuple | exception | error | jax, pytree, pickle |
| Malformed pickled SequenceKey, expected 1-tuple | exception | error | jax, pytree, pickle |
| Malformed pickled DictKey, expected 1-tuple | exception | error | jax, pytree, pickle |
| Malformed pickled GetAttrKey, expected 1-tuple | exception | error | jax, pytree, pickle |
| Malformed pickled FlattenedIndexKey, expected 1-tuple | exception | error | jax, pytree, pickle |
| Could not find memory addressable by device %s. Device %s ca | validation | error | jax, sharding, memory-kind, tpu |
| Got invalid memory kind: %s. Valid memory kinds are: %s | validation | error | jax, sharding, memory-kind |
| Invalid StrongLRUCache pickle version, got %d, expected 1 | validation | error | jax, lru-cache, pickle, versioning |
| Cannot enter TracebackScope recursively. | validation | error | jax, traceback, context-manager |
| code argument must be a code object | validation | error | jax, traceback, type-check |
| Invalid pipeline stage | validation | error | jax, xla, compiler-passes, validation |
| c_api argument to register_xla_transform_c_api is not a pjrt | validation | error | jax, xla, pjrt, pycapsule |
| Cannot register XLA transform '%s': PJRT plugin does not sup | exception | error | jax, xla, pjrt, plugin-capabilities |
| c_api argument to clear_xla_transform_c_api is not a pjrt_c_ | validation | error | jax, xla, pjrt, pycapsule |