ErrLookup › Lightning-AI/pytorch-lightning
Lightning-AI/pytorch-lightning
Pretrain, finetune ANY AI model of ANY size on 1 or 10,000+ GPUs with zero code changes. · Python · 270 source files
Analyzed at 9fed5c27d2 on 2026-08-28. 526 documented errors.
| Code / Message | Type | Severity | Tags |
|---|---|---|---|
| Device should be CPU, got {device} instead. | validation | error | pytorch-lightning, self-log, duplicate-metric, misconfiguration |
| `devices` selected with `CPUAccelerator` should be an int > | validation | error | pytorch-lightning, torchmetrics, compute, type-mismatch |
| Device should be CUDA, got {device} instead. | validation | error | pytorch-lightning, trainer, fast-dev-run, validation |
| You requested to find {num_devices} devices but there are no | validation | error | pytorch-lightning, trainer, limit-batches, validation |
| You requested to find {num_devices} devices but this machine | validation | error | pytorch-lightning, trainer, profiler, invalid-argument |
| You requested to find {num_devices} devices but only {len(av | error_code | error | pytorch-lightning, trainer, val-check-interval, time-format |
| Device should be MPS, got {device} instead. | validation | error | pytorch-lightning, trainer, val-check-interval, parse-error |
| `name` must be a str, found {name} | validation | error | pytorch-lightning, trainer, barebones, checkpointing |
| '{name}' is already present in the registry. HINT: Use `over | validation | error | pytorch-lightning, trainer, barebones, logger |
| '{}' not found in registry. Available names: {} | validation | error | pytorch-lightning, trainer, barebones, progress-bar |
| Received multiple values for {', '.join(duplicated_plugin_ke | validation | error | lightning, fabric, plugins, config-validation, duplicate-argument |
| Received both `precision={precision_input}` and `plugins={se | validation | error | lightning, fabric, precision, plugins, conflicting-arguments |
| accelerator set through both strategy class and accelerator | validation | error | lightning, fabric, strategy, accelerator, conflicting-arguments |
| precision set through both strategy class and plugins, choos | validation | error | lightning, fabric, strategy, precision, plugins, conflicting-arguments |
| checkpoint_io set through both strategy class and plugins, c | validation | error | lightning, fabric, strategy, checkpointio, plugins, conflicting-arguments |
| cluster_environment set through both strategy class and plug | validation | error | lightning, fabric, strategy, cluster-environment, plugins, conflicting-arguments |
| CPU parallel_devices set through {self._strategy_flag.__clas | validation | error | lightning, fabric, strategy, accelerator, device-mismatch, cpu |
| GPU parallel_devices set through {self._strategy_flag.__clas | validation | error | lightning, fabric, strategy, accelerator, device-mismatch, gpu |
| `num_nodes` must be a positive integer, but got {num_nodes}. | validation | error | lightning, fabric, num-nodes, argument-validation, distributed |
| `Fabric(devices={self._devices_flag!r})` value is not a vali | validation | error | lightning, fabric, devices, argument-validation, no-gpus |
| Precision {repr(precision)} is invalid. Allowed precision va | validation | error | lightning, fabric, precision, config-validation |
| No models were set up for backward. Did you forget to call ` | exception | critical | lightning, fabric, deepspeed, backward, setup-order |
| When using multiple models + deepspeed, please provide the m | validation | error | lightning, fabric, deepspeed, multi-model, backward |
| Only one of `clip_val` or `max_norm` can be set as this spec | validation | error | lightning, fabric, gradient-clipping, mutually-exclusive-args |
| You have to specify either `clip_val` or `max_norm` to do gr | validation | error | lightning, fabric, gradient-clipping, missing-argument |
| You need to set up the model first before you can call `fabr | exception | error | lightning, fabric, no-backward-sync, ddp, setup-order |
| Filter should be a dictionary, given {filter!r} | validation | error | lightning, fabric, checkpointing, type-validation |
| The filter keys {filter.keys() - state} are not present in t | validation | error | lightning, fabric, checkpointing, key-mismatch |
| Expected `fabric.save(filter=...)` for key {k!r} to be a cal | validation | error | lightning, fabric, checkpointing, type-validation |
| This script was launched through the CLI, and processes have | exception | error | lightning, fabric, cli, launch, double-launch |
| `Fabric.launch(...)` needs to be a callable, but got {functi | validation | error | lightning, fabric, launch, callable, typo |
| `Fabric.launch(function={function})` needs to take at least | validation | error | lightning, fabric, launch, signature |
| To spawn processes with the `{type(self.strategy).__name__}` | validation | error | lightning, fabric, ddp-spawn, launch, xla |
| Overriding `Fabric.run()` and launching from the CLI is not | validation | error | lightning, fabric, cli, subclass, method-override |
| To use Fabric with more than one device, you must call `.lau | exception | critical | lightning, fabric, distributed, launch, ddp |
| A model should be passed only once to the `setup` method. | validation | error | lightning, fabric, setup, double-setup |
| An optimizer should be passed only once to the `setup` metho | validation | error | lightning, fabric, setup, optimizer, double-setup |
| The optimizer has references to the model's meta-device para | exception | error | lightning, fabric, fsdp, meta-device, init-module |
| A model should be passed only once to the `setup_module` met | validation | error | lightning, fabric, setup-module, double-setup |
| The `{type(self._strategy).__name__}` requires the model and | exception | error | lightning, fabric, deepspeed, xla, setup, joint-setup |
| `setup_optimizers` requires at least one optimizer as input. | validation | error | pytorch-lightning, fabric, optimizer, validation |
| An optimizer should be passed only once to the `setup_optimi | validation | error | pytorch-lightning, fabric, optimizer, double-setup |
| The optimizer has references to the model's meta-device para | exception | error | pytorch-lightning, fabric, meta-device, fsdp, lazy-init |
| `setup_dataloaders` requires at least one dataloader as inpu | validation | error | pytorch-lightning, fabric, dataloader, validation |
| A dataloader should be passed only once to the `setup_datalo | validation | error | pytorch-lightning, fabric, dataloader, double-setup |
| Only PyTorch DataLoader are currently supported in `setup_da | validation | error | pytorch-lightning, fabric, dataloader, type-validation |
| The `CSVLogger` does not yet support logging hyperparameters | exception | warning | pytorch-lightning, fabric, csv-logger, hyperparameters, not-implemented |
| Neither `tensorboard` nor `tensorboardX` is available. Try ` | exception | error | pytorch-lightning, tensorboard, missing-dependency, logger |
| you tried to log {v} which is currently not supported. Try a | validation | error | pytorch-lightning, tensorboard, logging, type-validation |
| `{type(self).__name__}` does not own a group. HINT: try `col | exception | error | pytorch-lightning, distributed, process-group, collectives |
| `{type(self).__name__}` already owns a group. | exception | error | pytorch-lightning, distributed, process-group, collectives, double-setup |
| `{type(self).__name__}` does not own a group to destroy. | exception | error | pytorch-lightning, distributed, process-group, collectives, teardown |
| Torch distributed is not available. | exception | critical | pytorch-lightning, distributed, pytorch-build, platform-support |
| Unsupported op {op!r} of type {type(op).__name__} | validation | error | pytorch-lightning, distributed, reduceop, type-validation |
| op {op!r} is not a member of `ReduceOp` | validation | error | pytorch-lightning, distributed, reduceop, invalid-value |
| The Kubeflow environment can't be detected automatically. | exception | error | pytorch-lightning, kubeflow, cluster-environment, distributed |
| Cannot determine world size. Environment variable `JSM_NAMES | validation | error | lsf, jsrun, hpc, missing-env-var, distributed |
| Cannot determine global rank. Environment variable `JSM_NAME | validation | error | lsf, jsrun, hpc, missing-env-var, distributed |
| Cannot determine local rank. Environment variable `JSM_NAMES | validation | error | lsf, jsrun, hpc, missing-env-var, distributed |
| Did not find the environment variable `LSB_DJOB_RANKFILE` | validation | error | lsf, bsub, hpc, missing-env-var, distributed |
| str(_XLA_AVAILABLE) | exception | error | xla, tpu, dependency, pytorch-lightning |
| `Trainer.save_checkpoint(..., storage_options=...)` with `st | validation | error | xla, checkpoint, storage-options, pytorch-lightning |
| Passed `{type(self).__name__}(precision={precision!r})`. Pre | validation | error | amp, precision, validation, pytorch-lightning |
| `precision='bf16-mixed'` does not use a scaler, found {scale | validation | error | amp, precision, gradscaler, pytorch-lightning |
| AMP and the LBFGS optimizer are not compatible. | validation | error | amp, lbfgs, optimizer, pytorch-lightning |
| Gradient clipping is not implemented for optimizers handling | exception | error | amp, gradient-clipping, fused-optimizer, pytorch-lightning |
| {mode!r} only works with `dtype=torch.float16`, but you chos | validation | error | bitsandbytes, quantization, int8, dtype, pytorch-lightning |
| You are using the bitsandbytes precision plugin, but your mo | validation | error | bitsandbytes, quantization, model-structure, pytorch-lightning |
| Instantiating your model under the `init_module` context man | validation | error | bitsandbytes, quantization, init-module, pytorch-lightning |
| str(_BITSANDBYTES_AVAILABLE) | exception | error | bitsandbytes, dependency, quantization, pytorch-lightning |
| `precision={precision!r})` is not supported in DeepSpeed. `p | validation | error | deepspeed, precision, config-validation, pytorch-lightning |
| `precision={precision!r})` is not supported in FSDP. `precis | validation | error | fsdp, precision, config-validation, pytorch-lightning |
| `precision={precision!r}` does not use a scaler, found {scal | validation | error | fsdp, precision, grad-scaler, pytorch-lightning |
| Gradient clipping is not implemented for optimizers handling | exception | error | fsdp, grad-scaler, optimizer-incompatibility, pytorch-lightning |
| str(_TRANSFORMER_ENGINE_AVAILABLE) | exception | error | transformer-engine, fp8, missing-dependency, gpu, pytorch-lightning |
| str(_XLA_AVAILABLE) | exception | error | xla, tpu, missing-dependency, pytorch-lightning |
| `precision={precision!r})` is not supported in XLA. `precisi | validation | error | xla, precision, config-validation, pytorch-lightning |
| Blocking backward sync is only possible if the module passed | validation | error | ddp, gradient-accumulation, type-mismatch, pytorch-lightning |
| To use the `DeepSpeedStrategy`, you must have DeepSpeed inst | exception | critical | deepspeed, missing-dependency, import-error, pytorch-lightning |
| PyTorch >= 2.6 requires DeepSpeed >= 0.16.0. Detected DeepSp | exception | critical | deepspeed, version-mismatch, pytorch-upgrade, pytorch-lightning |
| Currently only one optimizer is supported with DeepSpeed. Go | validation | error | deepspeed, multiple-optimizers, unsupported-operation, pytorch-lightning |
| The `{type(self).__name__}` does not support setting up the | validation | error | deepspeed, api-misuse, optimizer-setup, pytorch-lightning |
| `{empty_init=}` is not a valid choice with `DeepSpeedStrateg | validation | error | deepspeed, zero-3, model-init, pytorch-lightning |
| `DeepSpeedStrategy.save_checkpoint(..., storage_options=...) | validation | error | deepspeed, checkpointing, unsupported-argument, pytorch-lightning |
| `DeepSpeedStrategy.save_checkpoint(..., filter=...)` is not | validation | error | deepspeed, checkpointing, unsupported-argument, pytorch-lightning |
| Could not find a DeepSpeed model in the provided checkpoint | validation | error | deepspeed, checkpointing, state-validation, pytorch-lightning |
| Found multiple DeepSpeed engine modules in the given state. | validation | error | deepspeed, checkpointing, multiple-models, pytorch-lightning |
| Got DeepSpeedStrategy.load_checkpoint(..., state={state!r}) | validation | error | deepspeed, checkpointing, resume, state-validation, pytorch-lightning |
| Could not find a DeepSpeed model in the provided checkpoint | validation | error | deepspeed, checkpointing, resume, state-validation, pytorch-lightning |
| Found multiple DeepSpeed engine modules in the given state. | validation | error | deepspeed, checkpointing, resume, multiple-models, pytorch-lightning |
| DeepSpeed was unable to load the checkpoint. Ensure you pass | error_code | critical | deepspeed, checkpoint, load |
| DeepSpeed handles gradient clipping automatically within the | validation | error | deepspeed, gradient-clipping, not-implemented |
| The DeepSpeed strategy is only supported on CUDA GPUs but `{ | validation | critical | deepspeed, accelerator, cuda, gpu-required |
| To use DeepSpeed you must pass in a DeepSpeed config dict, o | validation | error | deepspeed, config, missing-argument |
| You passed in a path to a DeepSpeed config but the path does | error_code | error | deepspeed, config, file-not-found, path |
| The selected device indices {selected_device_indices!r} don' | error_code | critical | deepspeed, cuda, device-selection, multi-gpu |
| {default_message}. It looks like you passed the path to a su | error_code | error | deepspeed, checkpoint, path, load |
| {default_message}. It looks like you passed the path to a fi | error_code | error | deepspeed, checkpoint, path, load |
| The provided path is not a valid DeepSpeed checkpoint: {path | error_code | error | deepspeed, checkpoint, path, load |
| The `{type(self).__name__}` does not use the `CheckpointIO` | validation | error | fsdp, checkpoint, not-implemented, api-misuse |
| The `{type(self).__name__}` does not support setting a `Chec | validation | error | fsdp, checkpoint, not-implemented, setter |
| The FSDP strategy can only work with the `FSDPPrecision` plu | validation | error | fsdp, precision, type-error |
| You set `{type(self).__name__}(use_orig_params=False)` but t | validation | error | fsdp, use-orig-params, optimizer, setup |
| The optimizer does not seem to reference any FSDP parameters | validation | error | fsdp, optimizer, flat-params, setup-order |
| Gradient clipping with FSDP is only possible if the module p | validation | error | fsdp, gradient-clipping, type-error, wrapping |
| `FSDPStrategy.save_checkpoint(..., storage_options=...)` is | validation | error | fsdp, checkpoint, storage-options, save |
| Could not find a FSDP model in the provided checkpoint state | validation | error | fsdp, checkpoint, state, save |
| Found multiple FSDP models in the given state. Saving checkp | validation | error | fsdp, checkpoint, save, pytorch-lightning, distributed |
| Unknown state_dict_type: {self._state_dict_type} | validation | error | fsdp, internal, state-dict, pytorch-lightning |
| Got FSDPStrategy.load_checkpoint(..., state={state!r}) but a | validation | error | fsdp, checkpoint, load, validation, pytorch-lightning |
| Loading a single optimizer object from a checkpoint is not s | validation | error | fsdp, optimizer, checkpoint, not-implemented, pytorch-lightning |
| Could not find a FSDP model in the provided checkpoint state | validation | error | fsdp, checkpoint, load, setup, pytorch-lightning |
| Found multiple FSDP models in the given state. Loading check | validation | error | fsdp, checkpoint, load, distributed, pytorch-lightning |
| The path {str(path)!r} does not point to a valid checkpoint. | error_code | error | fsdp, checkpoint, load, file-validation, distributed |
| You cannot set both `activation_checkpointing` and `activati | validation | error | fsdp, activation-checkpointing, config, mutually-exclusive, pytorch-lightning |
| The hybrid sharding strategy requires you to pass at least o | exception | error | pytorch-lightning, fabric, fsdp, distributed, sharding |
| The start method '{self._start_method}' is not available on | validation | error | pytorch-lightning, multiprocessing, platform, start-method, windows |
| Cannot re-initialize CUDA in forked subprocess. To use CUDA | exception | critical | pytorch-lightning, cuda, multiprocessing, fork, spawn |
| Lightning can't create new processes if CUDA is already init | exception | critical | pytorch-lightning, cuda, multiprocessing, pytorch-version |
| Launching multiple processes with the 'spawn' start method r | exception | error | pytorch-lightning, multiprocessing, spawn, entry-point, guard |
| The launcher can only create subprocesses once. | exception | error | pytorch-lightning, subprocess, launcher, ddp, lifecycle |
| Lightning attempted to launch new distributed processes with | exception | error | pytorch-lightning, local-rank, environment-variables, distributed, launcher |
| {str(_XLA_AVAILABLE)} | exception | error | pytorch-lightning, xla, tpu, missing-dependency, torch-xla |
| Accessing the device mesh before processes have initialized | exception | error | pytorch-lightning, model-parallel, device-mesh, initialization-order |
| The `{type(self).__name__}` does not use the `CheckpointIO` | exception | error | pytorch-lightning, model-parallel, checkpoint-io, not-implemented, api-contract |
| Could not find a distributed model in the provided checkpoin | validation | error | lightning, fabric, model-parallel, checkpoint, dtensor |
| Found multiple distributed models in the given state. Loadin | validation | error | lightning, fabric, model-parallel, checkpoint, multiple-models |
| The path {str(path)!r} does not point to a valid checkpoint. | validation | error | lightning, fabric, checkpoint, invalid-path |
| The sizes `data_parallel_size={data_parallel_size}` and `ten | exception | error | lightning, fabric, model-parallel, device-mesh, world-size |
| Failed to load checkpoint directly into the model. The given | validation | error | lightning, fabric, checkpoint, full-checkpoint, shards |
| The model contains a key '{full_param_name}' that does not e | validation | error | lightning, fabric, checkpoint, state-dict, strict-loading |
| `name` must be a str, found {name} | validation | error | lightning, fabric, registry, type-error |
| '{name}' is already present in the registry. HINT: Use `over | validation | error | lightning, fabric, registry, duplicate-registration |
| '{}' not found in registry. Available names: {} | exception | error | lightning, fabric, registry, unknown-strategy, key-error |
| {str(_XLA_AVAILABLE)} | exception | error | lightning, fabric, xla, tpu, missing-dependency |
| Could not find a XLAFSDP model in the provided checkpoint st | exception | error | xla, fsdp, checkpoint, save, model-wrapping, lightning-fabric |
| Found multiple XLAFSDP modules in the given state. Saving ch | exception | error | xla, fsdp, checkpoint, multiple-models, lightning-fabric |
| Multihost setups do not have a shared filesystem, so the che | exception | error | xla, fsdp, tpu, multihost, checkpoint, consolidation, lightning-fabric |
| Got `XLAFSDPStrategy.load_checkpoint(..., state={state!r})` | exception | error | xla, fsdp, checkpoint, load, state-required, lightning-fabric |
| Loading a single module or optimizer object from a checkpoin | exception | error | xla, fsdp, checkpoint, load, not-implemented, lightning-fabric |
| The path {str(file)!r} does not point to valid sharded check | exception | error | xla, fsdp, checkpoint, load, sharded, file-not-found, lightning-fabric |
| Could not find a XLAFSDP model in the provided checkpoint st | exception | error | xla, fsdp, checkpoint, load, model-wrapping, lightning-fabric |
| Found multiple XLAFSDP modules in the given state. Loading c | exception | error | xla, fsdp, checkpoint, load, multiple-models, lightning-fabric |
| The path {str(path)!r} does not point to a valid full checkp | exception | error | xla, fsdp, checkpoint, load, full-checkpoint, file-not-found, lightning-fabric |
| Found a XLAFSDP model in the provided checkpoint state. Plea | exception | error | xla, fsdp, checkpoint, load, full-checkpoint, model-wrapping, lightning-fabric |
| XLAFSDP only supports a single model instance with 'model' a | exception | error | xla, fsdp, checkpoint, load, full-checkpoint, not-implemented, lightning-fabric |
| Unknown state_dict_type: {self._state_dict_type} | exception | error | xla, fsdp, config, invalid-value, lightning-fabric |
| You cannot set both `auto_wrapper_callable` and `activation_ | exception | error | xla, fsdp, activation-checkpointing, config-conflict, lightning-fabric |
| `activation_checkpointing_policy` must be a set, found {poli | exception | error | xla, fsdp, activation-checkpointing, type-error, lightning-fabric |
| Blocking backward sync is only possible if the module passed | exception | error | xla, fsdp, gradient-accumulation, no-sync, type-error, lightning-fabric |
| The metric `{value}` does not contain a single element, thus | exception | error | metrics, tensor, scalar, logging, value-error, lightning-fabric |
| Upgrade fsspec to enable cross-device local checkpoints: pip | exception | error | fsspec, checkpoint, cross-device, exdev, atomic-save, environment, lightning-fabric |
| Remote (fsspec) distributed checkpoints require `torch.distr | exception | error | distributed-checkpoint, fsspec, remote-storage, import-error, pytorch-version, lightning-fabric |
| The dataloader {dataloader} needs to subclass `torch.utils.d | exception | error | dataloader, distributed, sampler, type-error, lightning-fabric |
| Trying to inject custom `Sampler` into the `{dataloader_cls_ | exception | error | dataloader, distributed, sampler, misconfiguration, lightning-fabric |
| Trying to inject parameters into the `{dataloader_cls_name}` | exception | error | pytorch-lightning, dataloader, distributed, kwargs, init-signature |
| Trying to inject a modified sampler into the batch sampler; | exception | error | pytorch-lightning, batch-sampler, distributed, ddp |
| Lightning can't inject a (distributed) sampler into your ba | exception | error | pytorch-lightning, batch-sampler, distributed-sampler, ddp |
| The {constructor.__name__} implementation has an error where | exception | error | pytorch-lightning, kwargs, duplicate-argument, dataloader |
| `local_world_size` should be >= 1, got {local_world_size}. | exception | error | pytorch-lightning, validation, num-workers, world-size |
| '{type(self).__name__}' object has no attribute '{key}' | exception | error | pytorch-lightning, attribute-dict, config, keyerror |
| Cannot set the dtype explicitly. Please use module.to(new_dt | exception | error | pytorch-lightning, dtype, module, property-setter |
| GPUs should be a list | exception | error | pytorch-lightning, gpu, device-parser, internal-api |
| GPUs requested but none are available. | exception | critical | pytorch-lightning, cuda, gpu, environment, device-availability |
| At least one gpu type should be specified! | exception | error | gpu, device-configuration, lightning, cuda, mps |
| You requested gpu: {gpus} But your machine only has: {all_a | exception | error | gpu, device-configuration, cuda, multi-node, lightning |
| Device ID's (GPU) must be unique. | exception | error | gpu, duplicate-ids, device-configuration, validation |
| Device IDs (GPU/TPU) must be an int, a string, a sequence of | validation | error | devices, typeerror, null-config, lightning |
| Device IDs (GPU/TPU) must be an int, a string, a sequence of | validation | error | devices, typeerror, numpy, sequence-validation |
| Device IDs (GPU/TPU) must be an int, a string, a sequence of | validation | error | devices, typeerror, config, validation |
| Unable to determine if the path belongs to a shared filesyst | exception | error | distributed, filesystem, checkpointing, shared-storage |
| torch.distributed is not available. Cannot initialize distri | exception | critical | distributed, pytorch, environment, runtimeerror |
| You seem to have configured a sampler in your DataLoader whi | validation | error | dataloader, sampler, distributed-training, lightning |
| You seem to have configured a sampler in your DataLoader whi | validation | error | dataloader, sampler, infinite-iterator, distributed-training |
| `num_processes` should be >= 1, got {num_processes}. | validation | error | lightning, multiprocessing, configuration |
| Materialization requires that the `{type(module).__name__}.r | validation | error | lightning, meta-device, model-initialization |
| Expected `torch.nn.Module` or `torch.optim.Optimizer`, got: | validation | error | lightning, type-error, setup |
| '{type(self).__name__}' object has no attribute '{name}' | exception | error | lightning, lazy-loading, checkpoints |
| Path {str(filename)!r} does not exist or is not a file. | exception | error | lightning, checkpoints, file-not-found |
| Invalid seed specified via PL_GLOBAL_SEED: {repr(env_seed)} | validation | error | lightning, seeding, environment-variables |
| {seed} is not in bounds, numpy accepts from {min_seed_value} | validation | error | lightning, seeding, validation |
| SpikeDetection requires `torchmetrics>=1.0.0` Please upgrade | exception | error | lightning, torchmetrics, version-mismatch |
| Invalid mode. Has to be min or max, found {self.mode} | validation | error | lightning, spike-detection, configuration |
| Expected samples ({samples}) to be greater or equal than bat | validation | error | lightning, throughput, validation |
| Expected lengths ({lengths}) to be greater or equal than sam | validation | error | pytorch-lightning, throughput, validation, argument-mismatch |
| If lengths are passed ({len(self._lengths)}), there needs to | exception | error | pytorch-lightning, throughput, state-consistency |
| Expected a precision plugin, got {plugin} | exception | error | pytorch-lightning, fabric, precision, type-mismatch |
| Expected the value to increase, last: {last}, current: {x} | validation | error | pytorch-lightning, throughput, monotonic, timing |
| __setitem__ is not supported | exception | warning | pytorch-lightning, not-implemented, internal-api |
| Expected a method or a string, but got: {type(method).__name | validation | error | pytorch-lightning, fabric, type-mismatch, method-binding |
| You cannot mark the forward method itself as a forward metho | validation | error | pytorch-lightning, fabric, invalid-argument |
| You marked '{name}' as a forward method, but `{type(self._or | exception | error | pytorch-lightning, fabric, attribute-error, method-lookup |
| You are calling the method `{type(self._original_module).__n | exception | error | pytorch-lightning, fabric, ddp, gradient-sync, method-bypass |
| Failed to determine the arguments that were used to compile | exception | error | pytorch-lightning, torch-compile, import-order |
| `mode` can be {', '.join(self.mode_dict.keys())}, got {self. | validation | error | lightning, early-stopping, callback-config, invalid-argument |
| Early stopping conditioned on metric `{self.monitor}` which | exception | error | lightning, early-stopping, metric-not-found, monitor |
| The Finetuning callback does not support running with the De | exception | error | lightning, finetuning, deepspeed, strategy-incompatible |
| The LightningModule should have a nn.Module `backbone` attri | validation | error | lightning, finetuning, backbone, attribute-missing |
| Empty dict cannot be interpreted correct | validation | error | lightning, gradient-accumulation, empty-argument, callback-config |
| Epoch should be an int greater than or equal to 0. Got {list | validation | error | lightning, gradient-accumulation, invalid-key, type-validation |
| Accumulation factor should be an int greater than 0. Got {li | validation | error | lightning, gradient-accumulation, invalid-value, type-validation |
| Epochs indexing from 1, epoch {minimal_epoch} cannot be inte | validation | error | lightning, gradient-accumulation, unreachable, defensive-code |
| Automatic gradient accumulation and the `GradientAccumulatio | exception | error | lightning, gradient-accumulation, manual-optimization, incompatible-config |
| The `{type(trainer.strategy).__name__}` does not support `ac | exception | error | lightning, deepspeed, gradient-accumulation, strategy-incompatible |
| You have set `accumulate_grad_batches` and are using the `Gr | validation | error | lightning, gradient-accumulation, config-conflict, trainer-options |
| `mode` should be either of {self.SUPPORTED_MODES} | validation | error | lightning, lr-finder, invalid-argument, hyperparameter-tuning |
| logging_interval should be `step` or `epoch` or `None`. | validation | error | lightning, lr-monitor, invalid-argument, callback-config |
| Cannot use `LearningRateMonitor` callback with `Trainer` tha | validation | error | lightning, lr-monitor, missing-logger, trainer-config |
| A single `Optimizer` cannot have multiple parameter groups w | validation | error | lightning, lr-monitor, optimizer, duplicate-name |
| `ModelCheckpoint(save_last='link')` is only supported for lo | validation | error | lightning, model-checkpoint, remote-storage, symlink |
| `ModelCheckpoint(monitor={self.monitor!r})` could not find t | validation | error | lightning, model-checkpoint, metric-not-found, monitor |
| Invalid value for save_top_k={self.save_top_k}. Must be >= - | validation | error | lightning, model-checkpoint, invalid-argument, callback-config |
| Invalid value for every_n_train_steps={self._every_n_train_s | validation | error | pytorch-lightning, modelcheckpoint, config-validation, argument-validation |
| Invalid value for every_n_epochs={self._every_n_epochs}. Mus | validation | error | pytorch-lightning, modelcheckpoint, config-validation, argument-validation |
| Combination of parameters every_n_train_steps={self._every_n | validation | error | pytorch-lightning, modelcheckpoint, mutually-exclusive-args, config-validation |
| ModelCheckpoint(save_top_k={self.save_top_k}, monitor=None) | validation | error | pytorch-lightning, modelcheckpoint, monitor, config-validation |
| `mode` can be {', '.join(mode_dict.keys())} but got {mode} | exception | error | pytorch-lightning, modelcheckpoint, mode, argument-validation |
| The filename cannot be empty | exception | error | pytorch-lightning, on-exception-checkpoint, filename, argument-validation |
| `write_interval` should be one of {[i.value for i in WriteIn | exception | error | pytorch-lightning, prediction-writer, write-interval, argument-validation |
| The `PredictionWriterCallback` does not support using `datal | exception | error | pytorch-lightning, prediction-writer, dataloader-iter, unsupported-operation |
| `RichProgressBar` requires `rich` >= 10.2.2. Install it by r | exception | error | pytorch-lightning, rich, missing-dependency, progress-bar |
| The provided `parameter_names` name: {name} isn't in {self.P | exception | error | pytorch-lightning, pruning, parameter-names, argument-validation |
| The provided `pruning_fn` {pruning_fn} isn't available in Py | exception | error | pytorch-lightning, pruning, pruning-fn, argument-validation |
| When requesting `structured` pruning, the `pruning_dim` shou | exception | error | pytorch-lightning, pruning, structured-pruning, missing-argument |
| When requesting `ln_structured` pruning, the `pruning_norm` | exception | error | pytorch-lightning, pruning, ln-structured, missing-argument |
| PyTorch `BasePruningMethod` is currently only supported with | exception | error | pytorch-lightning, pruning, base-pruning-method, unsupported-operation |
| `pruning_fn` is expected to be a str in {list(_PYTORCH_PRUNI | exception | error | pytorch-lightning, pruning, pruning-fn, type-error |
| Only the "unstructured" PRUNING_TYPE is supported with `use_ | exception | error | pytorch-lightning, pruning, global-unstructured, unsupported-operation |
| `amount` should be provided and be either an int, a float or | exception | error | pytorch-lightning, pruning, amount, type-error |
| `verbose` must be any of (0, 1, 2) | exception | error | pytorch-lightning, pruning, verbose, argument-validation |
| Some provided `parameters_to_prune` don't exist in the model | exception | error | pytorch-lightning, pruning, parameters-to-prune, model-mismatch |
| The provided `parameters_to_prune` should either be list of | exception | error | pytorch-lightning, pruning, parameters-to-prune, argument-validation |
| `RichModelSummary` requires `rich` to be installed. Install | exception | error | rich, optional-dependency, model-summary, callback |
| outputs have to be of type torch.Tensor or Mapping, got {typ | exception | error | spike-detection, training-step, typeerror, callback |
| swa_epoch_start should be a >0 integer or a float between 0 | exception | error | swa, stochastic-weight-averaging, validation, callback |
| The `swa_lrs` should a positive float, or a list of positive | exception | error | swa, swa-lrs, validation, learning-rate |
| The `avg_fn` should be callable. | exception | error | swa, avg-fn, validation, callback |
| device is expected to be a torch.device or a str. Found {dev | exception | error | swa, device, validation, callback |
| SWA does not currently support sharded models. | exception | error | swa, fsdp, deepspeed, sharding, incompatible-strategy |
| SWA currently works with 1 `optimizer`. | exception | error | swa, multiple-optimizers, configure-optimizers, callback |
| SWA currently not supported for more than 1 `lr_scheduler`. | exception | error | swa, lr-scheduler, multiple-schedulers, callback |
| SWA with `swa_epoch_start` as a float is not supported when | exception | error | swa, max-epochs, open-ended-training, callback |
| `Timer(duration={duration!r})` is not a valid duration. Expe | exception | error | timer, duration-format, validation, callback |
| Unsupported parameter value `Timer(interval={interval})`. Po | exception | error | timer, interval, enum-validation, callback |
| {_JSONARGPARSE_SIGNATURES_AVAILABLE} | exception | error | lightning-cli, jsonargparse, dependency-version, module-not-found |
| Cannot add arguments from: {lightning_class}. You should pro | exception | error | lightning-cli, add-arguments, type-validation, parser |
| `save_to_log_dir=False` only makes sense when subclassing Sa | exception | error | lightning-cli, save-config, save-to-log-dir, misuse-guard |
| {self.__class__.__name__} expected {config_path} to NOT exis | exception | error | lightning-cli, save-config, overwrite, file-exists, idempotent-rerun |
| `{self.__class__.__name__}.add_configure_optimizers_method_t | exception | error | lightning-cli, configure-optimizers, multiple-optimizers, automatic-mode |
| `train_dataloader` must be implemented to be used with the L | exception | error | train-dataloader, not-implemented, lightning-module, required-method |
| `test_dataloader` must be implemented to be used with the Li | exception | error | test-dataloader, not-implemented, lightning-module, required-method |
| `val_dataloader` must be implemented to be used with the Lig | exception | critical | pytorch-lightning, lightning-module, dataloader, validation, hook-not-implemented |
| `predict_dataloader` must be implemented to be used with the | exception | error | pytorch-lightning, lightning-module, dataloader, inference, hook-not-implemented |
| Primitives {_PRIMITIVE_TYPES} are not allowed. | exception | error | pytorch-lightning, hparams, save-hyperparameters, type-validation |
| Unsupported config type of {type(hp)}. | exception | error | pytorch-lightning, hparams, save-hyperparameters, type-validation, unsupported-type |
| {self.__class__.__qualname__} is not attached to a `Trainer` | exception | error | pytorch-lightning, lightning-module, trainer, lifecycle, not-attached |
| You are trying to `self.log()` but the loop's result collect | exception | error | pytorch-lightning, self-log, predict-step, logging, misconfiguration |
| You are trying to `self.log()` but it is not managed by the | exception | error | pytorch-lightning, self-log, control-flow, logging, misconfiguration |
| You called `self.log` with the key `{name}` but it should no | exception | error | pytorch-lightning, self-log, multi-dataloader, key-validation |
| Could not find the `LightningModule` attribute for the `torc | exception | error | pytorch-lightning, self-log, torchmetrics, metric-registration |
| Could not find the `LightningModule` attribute for the `torc | exception | error | pytorch-lightning, self-log, torchmetrics, metric-attribute |
| With `def training_step(self, dataloader_iter)`, `self.log(. | exception | error | pytorch-lightning, self-log, dataloader-iter, batch-size, training-step |
| `self.log_dict({dictionary})` was called, but nested diction | exception | error | pytorch-lightning, fabric, log-dict, nested-dict, validation |
| `self.log({name}, {value})` was called, but nested dictionar | exception | error | pytorch-lightning, fabric, self-log, nested-dict, validation |
| `self.log({name}, {value})` was called, but `{type(v).__name | exception | error | pytorch-lightning, fabric, self-log, type-validation, non-numeric |
| `self.log({name}, {value})` was called, but the tensor must | exception | error | pytorch-lightning, self-log, tensor, scalar, shape-validation |
| You have set `Trainer(gradient_clip_val={self.trainer.gradie | exception | error | pytorch-lightning, gradient-clipping, trainer-config, conflict |
| You have set `Trainer(gradient_clip_algorithm={self.trainer. | exception | error | pytorch-lightning, gradient-clipping, trainer-config, conflict |
| `gradient_clip_val` should be an int or a float. Got {gradie | exception | error | pytorch-lightning, gradient-clipping, type-error, validation |
| `gradient_clip_algorithm` {gradient_clip_algorithm} is inval | exception | error | pytorch-lightning, gradient-clipping, invalid-argument, enum-validation |
| to use {fn_name}, please disable automatic optimization: set | exception | error | pytorch-lightning, manual-optimization, manual-backward, misconfiguration |
| `{type(self).__name__}.to_onnx()` requires `onnx` to be inst | exception | error | onnx, export, missing-dependency, lightning |
| `{type(self).__name__}.to_onnx(dynamo=True)` requires `onnxs | exception | error | onnx, dynamo, onnxscript, missing-dependency, lightning |
| Could not export to ONNX since neither `input_sample` nor `m | exception | error | onnx, export, example-input, lightning |
| Choosing method=`trace` requires either `example_inputs` or | exception | error | torchscript, trace, example-input, lightning |
| The 'method' parameter only supports 'script' or 'trace', bu | exception | error | torchscript, invalid-argument, lightning |
| `{type(self).__name__}.to_tensorrt` requires `torch_tensorrt | exception | error | tensorrt, export, missing-dependency, lightning |
| TensorRT only supports CUDA devices. The current device is { | exception | error | tensorrt, cuda, device-mismatch, lightning |
| Could not export to TensorRT since neither `input_sample` no | exception | error | tensorrt, example-input, lightning |
| TensorRT with IR mode 'ts' only supports output format 'torc | exception | error | tensorrt, invalid-combination, torchscript, lightning |
| Your LightningModule code tried to access `self.trainer.{ite | exception | error | fabric, trainer, attribute-error, lightning |
| When `optimizer.step(closure)` is called, the closure should | exception | error | optimizer, closure, manual-optimization, lightning |
| Unknown configuration for model optimizers. Output from `mod | exception | critical | configure-optimizers, validation, lightning, misconfiguration |
| The lr scheduler dict must have the key "scheduler" with its | exception | critical | lr-scheduler, configure-optimizers, validation, lightning |
| The "interval" key in lr scheduler dict must be "step" or "e | exception | critical | lr-scheduler, interval, validation, lightning |
| `configure_optimizers` must include a monitor when a `Reduce | exception | critical | lr-scheduler, reduce-lr-on-plateau, monitor, lightning |
| The provided lr scheduler `{scheduler.__class__.__name__}` i | exception | error | lr-scheduler, checkpointing, type-error, lightning |
| The provided lr scheduler `{scheduler.__class__.__name__}` d | exception | error | lr-scheduler, custom-scheduler, hook, lightning |
| Training with multiple optimizers is only supported with man | exception | critical | optimizer, manual-optimization, migration, lightning |
| Training with multiple optimizers is only supported with man | exception | critical | optimizer, manual-optimization, gan, lightning |
| Some schedulers are attached with an optimizer that wasn't r | exception | critical | lr-scheduler, optimizer, validation, lightning |
| .csv, .yml or .yaml is required for `hparams_file` | exception | error | lightning, checkpoint, hparams, file-extension |
| Unsupported {cls} | exception | error | lightning, checkpoint, not-implemented, lightningmodule |
| The instantiator {instantiator_path!r} from the checkpoint i | exception | critical | lightning, checkpoint, security, arbitrary-code-execution, allowlist |
| You set `.load_from_checkpoint(..., strict={strict!r})` whic | exception | error | lightning, checkpoint, strict-loading, config-conflict |
| Missing folder: {os.path.dirname(tags_csv)}. | exception | error | lightning, hparams, csv, missing-directory, fsspec |
| Missing folder: {os.path.dirname(config_yaml)}. | exception | error | lightning, hparams, yaml, missing-directory, fsspec |
| hparams must be dictionary | exception | error | lightning, hparams, yaml, type-error, serialization |
| Experiment is not initialized | exception | error | litlogger, lightning, logger, lazy-initialization |
| `synchronous` requires mlflow>=2.8.0 | exception | error | mlflow, lightning, logger, version-mismatch, dependency |
| NeptuneLogger is no longer supported. Neptune has been sunse | exception | error | neptune, lightning, logger, removed-api, sunset |
| Error while merging hparams: the keys {inconsistent_keys} ar | exception | error | lightning, hparams, merge-conflict, datamodule |
| Providing log_model={log_model} and offline={offline} is an | exception | error | wandb, lightning, logger, offline, config-conflict |
| Expected a list as "images", found {type(images)} | exception | error | wandb, lightning, logging, images, type-error |
| Expected {n} items but only found {len(v)} for {k} | exception | error | wandb, lightning, logging, images, length-mismatch |
| Expected a list as "audios", found {type(audios)} | exception | error | wandb, lightning, logging, audio, type-error |
| Expected a list as "videos", found {type(videos)} | exception | error | wandb, lightning, logging, video, type-error |
| `{self.__class__.__name__}` should have been `setup` with a | exception | error | lightning, loops, fetchers, setup-order, internal-api |
| `prefetch_batches` should at least be 0. | exception | error | lightning, data-loading, prefetch, validation |
| `max_epochs` must be a non-negative integer or -1. You passe | exception | error | pytorch-lightning, trainer, max-epochs, configuration, validation |
| `val_check_interval` ({trainer.val_check_interval}) must be | exception | error | pytorch-lightning, validation, val-check-interval, scheduling |
| When using an IterableDataset for `train_dataloader`, `Train | exception | error | pytorch-lightning, iterable-dataset, streaming, validation |
| `{type(self).__name__}` does not support the `CombinedLoader | exception | error | pytorch-lightning, combined-loader, multi-dataloader, training |
| In automatic_optimization, when `training_step` returns a di | exception | error | pytorch-lightning, training-step, loss, automatic-optimization |
| In automatic optimization, `training_step` must return a Ten | exception | error | pytorch-lightning, training-step, return-type, automatic-optimization |
| Skipping the `training_step` by returning None in distribute | exception | critical | pytorch-lightning, distributed, ddp, training-step, sync |
| The closure hasn't been executed. HINT: did you call `optimi | exception | error | pytorch-lightning, optimizer-step, closure, automatic-optimization |
| In manual optimization, `training_step` must either return a | exception | error | pytorch-lightning, manual-optimization, training-step, return-type |
| `return_predictions` should be set to `False` when using the | exception | error | pytorch-lightning, predict, ddp-spawn, return-predictions |
| `trainer.predict()` only supports the `CombinedLoader(mode=" | exception | error | pytorch-lightning, predict, combined-loader, multi-dataloader |
| `max_steps` must be a non-negative integer or -1 (infinite s | exception | error | pytorch-lightning, trainer, max-steps, validation |
| ReduceLROnPlateau conditioned on metric {monitor_key} which | exception | error | pytorch-lightning, lr-scheduler, reducelronplateau, monitor-metric, logging |
| The loss returned in `training_step` is {loss}. | exception | critical | pytorch-lightning, nan-loss, numerical-stability, training |
| DataFetcher is unsupported for {trainer.state.stage} | exception | error | pytorch-lightning, internal, running-stage, data-fetcher |
| You provided only a single `{stage.dataloader_prefix}_datalo | exception | error | pytorch-lightning, dataloader-idx, hook-signature, single-dataloader |
| You provided multiple `{stage.dataloader_prefix}_dataloader` | exception | error | pytorch-lightning, dataloader-idx, hook-signature, multi-dataloader |
| The given dataset must implement the `__len__` method. | exception | error | pytorch-lightning, distributed, distributed-sampler, iterable-dataset |
| `Passed `{type(self).__name__}(precision={precision!r})`. Pr | exception | error | pytorch-lightning, precision, amp, mixed-precision, version-migration |
| `precision='bf16-mixed'` does not use a scaler, found {scale | exception | error | pytorch-lightning, amp, bf16, mixed-precision, gradscaler, plugin |
| AMP and the LBFGS optimizer are not compatible. | exception | error | pytorch-lightning, amp, lbfgs, optimizer, gradscaler, optimizer-step |
| The current optimizer, {type(optimizer).__qualname__}, does | exception | error | pytorch-lightning, amp, fused-optimizer, gradient-clipping, adamw, gradscaler |
| `Trainer(strategy='deepspeed', precision={precision!r})` is | exception | error | pytorch-lightning, deepspeed, precision, config-validation |
| DeepSpeed and the LBFGS optimizer are not compatible. | exception | error | pytorch-lightning, deepspeed, lbfgs, optimizer, distributed |
| Skipping backward by returning `None` from your `training_st | exception | error | pytorch-lightning, deepspeed, training-step, backward, automatic-optimization |
| `precision={precision!r})` is not supported in FSDP. `precis | exception | error | pytorch-lightning, fsdp, precision, config-validation, distributed |
| `precision={precision!r}` does not use a scaler, found {scal | exception | error | pytorch-lightning, fsdp, shardedgradscaler, mixed-precision, bf16 |
| `gradient_clip_algorithm='norm'` is currently not supported | exception | error | pytorch-lightning, fsdp, gradient-clipping, clip-norm, distributed |
| Was unable to infer precision type, received {self.precision | exception | error | pytorch-lightning, fsdp, mixed-precision, dtype-mapping, config-validation |
| {_XLA_AVAILABLE} | exception | critical | xla, tpu, dependency-missing, pytorch-lightning |
| `precision={precision!r})` is not supported in XLA. `precisi | exception | error | xla, precision, invalid-argument, pytorch-lightning |
| Skipping backward by returning `None` from your `training_st | exception | error | xla, training-step, backward, pytorch-lightning, misconfiguration |
| Attempting to stop recording an action ({action_name}) which | exception | error | profiler, advanced-profiler, api-misuse, pytorch-lightning |
| You are trying to use `ScheduleWrapper` which require kineto | exception | error | profiler, kineto, dependency-missing, pytorch |
| Found sort_by_key: {self._sort_by_key}. Should be within {se | exception | error | profiler, invalid-argument, pytorch-lightning |
| Found invalid table_kwargs key: {key}. This is already a pos | exception | error | profiler, invalid-argument, kwargs, pytorch-lightning |
| Found invalid table_kwargs key: {key}. Should be within {val | exception | error | profiler, invalid-argument, kwargs, version-compat |
| Schedule should be a callable. Found: {schedule} | exception | error | profiler, schedule, type-error, pytorch-lightning |
| Schedule should return a `torch.profiler.ProfilerAction`. Fo | exception | error | profiler, schedule, type-error, pytorch |
| The `configure_payload` method needs to be overridden. | exception | error | lightning, serving, not-implemented, pytorch |
| The `configure_serialization` method needs to be overridden. | exception | error | lightning, serving, serialization, not-implemented |
| The `serve_step` method needs to be overridden. | exception | error | lightning, serving, inference, not-implemented |
| The server didn't start within {self.timeout} seconds. | exception | error | lightning, serving, timeout, server-startup |
| Your provided payload {payload} should have a field named "b | exception | error | lightning, serving, payload, schema |
| The expected response {response} doesn't match the generated | exception | error | lightning, serving, response-mismatch, assertion |
| The model isn't servable. Investigate the traceback and try | exception | error | lightning, serving, misconfiguration, pipeline-gate |
| Please, return your outputs as a dictionary. Found {output} | exception | error | lightning, serving, return-type, contract |
| Post-localSGD algorithm is used, but model averaging period | exception | error | pytorch, lightning, ddp, distributed, post-local-sgd |
| Currently model averaging cannot work with a distributed opt | exception | error | pytorch, lightning, ddp, distributed-optimizer, incompatibility |
| To use the `DeepSpeedStrategy`, you must have DeepSpeed inst | exception | critical | deepspeed, strategy, dependencies, installation |
| PyTorch >= 2.6 requires DeepSpeed >= 0.16.0. Detected DeepSp | exception | critical | deepspeed, version-mismatch, pytorch-2-6, compatibility |
| The DeepSpeed strategy is only supported on CUDA GPUs but `{ | exception | critical | deepspeed, accelerator, cuda, hardware-requirement |
| Currently only one optimizer is supported with DeepSpeed. Go | exception | error | deepspeed, optimizers, multiple-optimizers, configure-optimizers |
| DeepSpeed does not support clipping gradients by value. | exception | error | deepspeed, gradient-clipping, trainer-flags |
| DeepSpeed currently only supports single optimizer, single o | exception | error | deepspeed, lr-scheduler, configure-optimizers |
| `{empty_init=}` is not a valid choice with `DeepSpeedStrateg | exception | error | deepspeed, zero-stage-3, weight-initialization |
| `Trainer.save_checkpoint(..., storage_options=...)` with `st | exception | error | deepspeed, save-checkpoint, storage-options, checkpointio |
| DeepSpeed was unable to load the checkpoint. Ensure you pass | exception | critical | deepspeed, load-checkpoint, resume, load-full-weights |
| You passed in a path to a DeepSpeed config but the path does | exception | error | deepspeed, config-file, file-not-found, paths |
| To use DeepSpeed you must pass in a DeepSpeed config dict, o | exception | critical | deepspeed, missing-config, zero-optimization |
| Do not set `gradient_accumulation_steps` in the DeepSpeed co | exception | error | deepspeed, gradient-accumulation, config-conflict |
| The FSDP strategy can only work with the `FSDPPrecision` plu | exception | error | fsdp, precision-plugin, mixed-precision, type-mismatch |
| The optimizer does not seem to reference any FSDP parameters | exception | critical | fsdp, optimizer, empty-parameter-list, use-orig-params |
| Unknown state_dict_type: {self._state_dict_type} | exception | error | fsdp, state-dict, invalid-argument, checkpointing |
| `FSDPStrategy.save_checkpoint(..., storage_options=...)` is | exception | error | fsdp, save-checkpoint, storage-options |
| The checkpoint path exists and is a directory: {path} | exception | error | fsdp, save-checkpoint, is-a-directory, paths |
| You have configured {len(self.optimizers)} optimizers but th | exception | critical | fsdp, resume, optimizer-states, checkpoint-mismatch |
| The path {str(path)!r} does not point to a valid checkpoint. | exception | error | fsdp, checkpoint, load-checkpoint, pytorch-lightning |
| The start method '{self._start_method}' is not available on | exception | error | multiprocessing, start-method, platform, pytorch-lightning |
| Calling `trainer.fit()` twice on the same Trainer instance u | exception | error | trainer, spawn, fit-twice, pytorch-lightning |
| The launcher can only create subprocesses once. | exception | error | subprocess, launcher, distributed, pytorch-lightning |
| Lightning attempted to launch new distributed processes with | exception | error | local-rank, distributed, subprocess, cluster-environment |
| raise ModuleNotFoundError(str(_XLA_AVAILABLE)) | exception | critical | xla, tpu, missing-dependency, pytorch-lightning |
| Calling `trainer.fit()` twice on the same Trainer instance u | exception | error | xla, tpu, fit-twice, trainer |
| Accessing the device mesh before processes have initialized | exception | error | device-mesh, model-parallel, lifecycle, pytorch-lightning |
| When using the {type(self).__name__}, you are required to ov | exception | error | model-parallel, fsdp2, configure-model, hook-required |
| Found modules that are wrapped with `torch.distributed.fsdp. | exception | error | fsdp, fsdp2, legacy-api, pytorch-version |
| `{type(self).__name__}.save_checkpoint(..., storage_options= | exception | error | checkpoint, storage-options, model-parallel, unsupported-argument |
| The checkpoint path exists and is a directory: {path} | exception | error | checkpoint, is-a-directory, model-parallel, path |
| raise ModuleNotFoundError(str(_XLA_AVAILABLE)) | exception | critical | xla, tpu, missing-dependency |
| The XLA strategy can only work with the `XLACheckpointIO` pl | exception | error | xla, checkpoint-io, plugin-mismatch |
| The XLA strategy can only work with the `XLAPrecision` plugi | exception | error | xla, precision-plugin, plugin-mismatch |
| raise ModuleNotFoundError(str(_XLA_AVAILABLE)) | exception | critical | xla, tpu, missing-dependency |
| The XLA strategy can only work with the `XLACheckpointIO` pl | exception | error | xla, checkpoint-io, plugin-mismatch |
| The XLA strategy can only work with the `XLAPrecision` plugi | exception | error | xla, precision-plugin, plugin-mismatch |
| Accessing the XLA device before processes have spawned is no | exception | error | xla, root-device, lifecycle |
| Currently, the XLAStrategy only supports `sum`, `mean`, `avg | exception | error | xla, reduce, reduce-op, distributed |
| No `{step_name}()` method defined to run `Trainer.{trainer_m | validation | error | pytorch-lightning, trainer, validation, missing-hook, configuration |
| Support for `{epoch_end_name}` has been removed in v2.0.0. ` | exception | error | pytorch-lightning, migration, v2-breaking-change, legacy-hooks |
| Automatic gradient clipping is not supported for manual opti | validation | error | pytorch-lightning, manual-optimization, gradient-clipping, configuration |
| Automatic gradient accumulation is not supported for manual | validation | error | pytorch-lightning, manual-optimization, gradient-accumulation, configuration |
| Both `{name}.configure_model`, and `{name}.configure_sharded | validation | error | pytorch-lightning, fsdp, deprecated-hook, configuration |
| You selected an invalid strategy name: `strategy={strategy!r | validation | error | pytorch-lightning, trainer, strategy, invalid-argument, ddp |
| You selected an invalid accelerator name: `accelerator={acce | validation | error | pytorch-lightning, trainer, accelerator, invalid-argument |
| You set `strategy={strategy}` but strategies from the DDP fa | validation | error | pytorch-lightning, mps, apple-silicon, ddp, strategy-mismatch |
| You set `Trainer(sync_batchnorm=True)` and provided a `{plug | validation | error | pytorch-lightning, sync-batchnorm, plugins, configuration-conflict |
| Found invalid type for plugin {plugin}. Expected one of: Pre | validation | error | pytorch-lightning, trainer, plugins, config-validation |
| Received multiple values for {', '.join(duplicated_plugin_ke | validation | error | pytorch-lightning, plugins, duplicate-config |
| Received both `precision={precision_flag}` and `plugins={sel | validation | error | pytorch-lightning, precision, conflicting-config |
| accelerator set through both strategy class and accelerator | validation | error | pytorch-lightning, strategy, accelerator, conflicting-config |
| precision set through both strategy class and plugins, choos | validation | error | pytorch-lightning, strategy, precision, conflicting-config |
| checkpoint_io set through both strategy class and plugins, c | validation | error | pytorch-lightning, strategy, checkpoint-io, conflicting-config |
| cluster_environment set through both strategy class and plug | validation | error | pytorch-lightning, strategy, cluster-environment, conflicting-config |
| CPU parallel_devices set through {self._strategy_flag.__clas | validation | error | pytorch-lightning, strategy, parallel-devices, device-mismatch |
| GPU parallel_devices set through {self._strategy_flag.__clas | validation | error | pytorch-lightning, strategy, parallel-devices, device-mismatch |
| `num_nodes` must be a positive integer, but got {num_nodes}. | validation | error | pytorch-lightning, num-nodes, input-validation |
| `Trainer(devices={self._devices_flag!r})` value is not a val | validation | error | pytorch-lightning, devices, config-validation |
| No supported gpu backend found! | validation | error | pytorch-lightning, gpu, cuda, environment |
| `{accelerator_cls.__qualname__}` can not run on your system | validation | error | pytorch-lightning, accelerator, hardware-availability |
| HPU is currently not supported. Please contact developer@lig | validation | error | pytorch-lightning, hpu, unsupported-feature |
| The strategy `{FSDPStrategy.strategy_name}` requires a GPU a | validation | error | pytorch-lightning, fsdp, strategy, cuda-only |
| You selected `Trainer(strategy='{strategy_flag}')` but proce | validation | error | pytorch-lightning, ddp-fork, windows, platform-unsupported |
| No precision set | exception | error | pytorch-lightning, precision, config-validation |
| Bitsandbytes is only supported on CUDA GPUs. | validation | error | pytorch-lightning, bitsandbytes, quantization, cuda-only |
| The `ModelParallelStrategy` does not support `Fabric(..., pr | validation | error | pytorch-lightning, model-parallel, precision, unsupported-combination |
| `Trainer(strategy={self._strategy_flag!r})` is not compatibl | validation | error | pytorch-lightning, ddp, jupyter, interactive-environment |
| The `XLAAccelerator` can only be used with a `SingleDeviceXL | validation | error | lightning, tpu, xla, strategy, trainer-init, accelerator |
| Trainer was configured with `enable_checkpointing=False` but | validation | error | lightning, trainer-init, checkpointing, callbacks, config-conflict |
| Unable to determine the source of the trainer. | exception | info | lightning, litmodels, inspect, dynamic-import, checkpointing |
| You added multiple progress bar callbacks to the Trainer, bu | validation | error | lightning, progress-bar, callbacks, trainer-init, duplicate-callbacks |
| Trainer was configured with `enable_progress_bar=False` but | validation | error | lightning, progress-bar, trainer-init, config-conflict, callbacks |
| Found more than one stateful callback of type `{type(callbac | validation | error | lightning, callbacks, state-dict, checkpointing, trainer-init |
| `.{fn}(ckpt_path="best")` is set but `ModelCheckpoint` is no | validation | error | lightning, checkpoint, best-model, validate, test, predict |
| You cannot execute `.{fn}(ckpt_path="best")` with `fast_dev_ | validation | error | lightning, fast-dev-run, checkpoint, best-model, smoke-test |
| `.{fn}(ckpt_path="best")` is set but `ModelCheckpoint` is no | validation | error | lightning, model-checkpoint, monitor, best-model, test |
| `.{fn}(ckpt_path="hpc")` is set but no HPC checkpoint was fo | validation | error | lightning, hpc, resume, checkpoint, slurm |
| `.{fn}()` found no path for the best weights: {ckpt_path!r}. | validation | error | lightning, checkpoint, resume, path-resolution, test |
| You restored a checkpoint with current_epoch={self.trainer.c | validation | error | lightning, resume, max-epochs, epoch, trainer-state |
| Trying to restore optimizer state but checkpoint contains on | validation | error | lightning, resume, optimizer-state, weights-only, checkpoint |
| Trying to restore learning rate scheduler state but checkpoi | validation | error | lightning, resume, lr-scheduler, weights-only, checkpoint |
| f"`check_val_every_n_epoch` should be an integer, found {che | validation | error | lightning, trainer-init, validation-interval, type-error, config |
| "`val_check_interval` should be an integer or a time-based d | validation | error | lightning, val-check-interval, validation, trainer-init, config |
| f"`reload_dataloaders_every_n_epochs` should be an int >= 0, | validation | error | lightning, trainer-init, dataloader-reload, type-error, config |
| f"`{hook_name}` is not a shared hook within `LightningModule | validation | error | lightning, internal-api, hooks, data-transfer, misuse |
| f"An invalid dataloader was passed to `Trainer.{trainer_fn.v | validation | error | lightning, dataloader, iterable, type-error, fit |
| f"An invalid dataloader was passed to `Trainer.{trainer_fn.v | validation | error | lightning, dataloader, iterable, lightning-module, hook-return |
| f"An invalid dataloader was returned from `{type(source.inst | validation | error | pytorch-lightning, dataloader, validation, typeerror |
| f"When using an `IterableDataset`, `Trainer(limit_{stage.dat | validation | error | pytorch-lightning, iterable-dataset, limit-batches, misconfiguration |
| f"You requested to check {limit_batches} of the `{stage.data | validation | error | pytorch-lightning, limit-batches, validation-loop, misconfiguration |
| f"Logging inside `{fx_name}` is not implemented." " Please, | exception | error | pytorch-lightning, logging, internal-error, runtimeerror |
| f"You can't `self.log()` inside `{fx_name}`. HINT: You can s | validation | error | pytorch-lightning, logging, hook-validation, misconfiguration |
| m.format("on_step", on_step, fx_name, fx_config["allowed_on_ | validation | error | pytorch-lightning, logging, on-step, misconfiguration |
| m.format("on_epoch", on_epoch, fx_name, fx_config["allowed_o | validation | error | pytorch-lightning, logging, on-epoch, misconfiguration |
| "`self.log(on_step=False, on_epoch=False)` is not useful." | validation | error | pytorch-lightning, logging, aggregation, misconfiguration |
| error | validation | error | pytorch-lightning, logging, reduce-fx, misconfiguration |
| f"`Trainer(barebones=True, log_every_n_steps={log_every_n_st | validation | error | trainer, barebones, logging, misconfiguration, pytorch-lightning |
| f"`Trainer(barebones=True, enable_model_summary={enable_mode | validation | error | trainer, barebones, model-summary, misconfiguration, pytorch-lightning |
| f"`Trainer(barebones=True, num_sanity_val_steps={num_sanity_ | validation | error | trainer, barebones, sanity-check, misconfiguration, pytorch-lightning |
| f"`Trainer(barebones=True, fast_dev_run={fast_dev_run!r})` w | validation | error | trainer, barebones, fast-dev-run, misconfiguration, pytorch-lightning |
| f"`Trainer(barebones=True, detect_anomaly={detect_anomaly!r} | validation | error | trainer, barebones, anomaly-detection, misconfiguration, pytorch-lightning |
| f"`Trainer(barebones=True, profiler={profiler!r})` was passe | validation | error | trainer, barebones, profiler, misconfiguration, pytorch-lightning |
| f"`gradient_clip_val` should be an int or a float. Got {grad | validation | error | trainer, gradient-clipping, type-error, hyperparameters, pytorch-lightning |
| f"`gradient_clip_algorithm` {gradient_clip_algorithm} is inv | validation | error | trainer, gradient-clipping, invalid-argument, misconfiguration, pytorch-lightning |
| "You cannot pass `train_dataloader` or `val_dataloaders` to | validation | error | trainer, fit, datamodule, dataloader, misconfiguration, pytorch-lightning |
| "`Trainer.validate()` requires a `LightningModule` when it h | validation | error | trainer, validate, missing-model, pytorch-lightning |
| "You cannot pass both `trainer.validate(dataloaders=..., dat | validation | error | trainer, validate, datamodule, dataloader, misconfiguration, pytorch-lightning |
| "`Trainer.test()` requires a `LightningModule` when it hasn' | validation | error | trainer, test, missing-model, checkpoint, pytorch-lightning |
| You cannot pass both `trainer.test(dataloaders=..., datamodu | validation | error | trainer, test, datamodule, dataloader, misconfiguration, pytorch-lightning |
| `Trainer.predict()` requires a `LightningModule` when it has | validation | error | trainer, predict, inference, missing-model, pytorch-lightning |
| You cannot pass both `trainer.predict(dataloaders=..., datam | validation | error | trainer, predict, datamodule, dataloader, misconfiguration, pytorch-lightning |
| Saving a checkpoint is only possible if a model is attached | validation | error | trainer, checkpoint, save, no-model-attached, pytorch-lightning |
| The attribute name for the learning rate was set to {attr_na | validation | error | lr-finder, tuner, attr-name, hyperparameters, pytorch-lightning |
| When using the learning rate finder, either `model` or `mode | validation | error | lr-finder, tuner, learning-rate, hyperparameters, pytorch-lightning |
| `model.configure_optimizers()` returned {len(optimizers)}, b | validation | error | lr-finder, tuner, multiple-optimizers, configure-optimizers, pytorch-lightning |
| To use the `plot` method, you must have Matplotlib installed | validation | warning | lr-finder, plot, matplotlib, missing-dependency, optional-install, pytorch-lightning |
| method='fit' is the only valid configuration to run lr finde | validation | error | lr-finder, tuner, misconfiguration, lightning |
| method {method!r} is invalid. Should be one of {supported_me | validation | error | tuner, validation, method-argument |
| In tuner with method={method!r}, `dataloaders` argument shou | validation | error | tuner, dataloaders, misconfiguration |
| In tuner with `method`={method!r}, `train_dataloaders` and ` | validation | error | tuner, dataloaders, misconfiguration |
| Trainer is already configured with a `LearningRateFinder` ca | validation | error | lr-finder, callbacks, tuner, duplicate-config |
| Tuning the batch size is currently not supported with distri | validation | error | batch-size-finder, distributed, ddp, tuner |
| Trainer is already configured with a `BatchSizeFinder` callb | validation | error | batch-size-finder, callbacks, tuner, duplicate-config |
| Mismatch in number of limits ({len(limits)}) and number of i | validation | error | combined-loader, limits, validation |
| Unsupported mode {mode!r}, please select one of: {list(_SUPP | validation | error | combined-loader, mode, invalid-argument |
| Mismatch in flattened length ({len(flattened)}) and existing | validation | error | combined-loader, mutation, length-mismatch |
| Mismatch in number of limits ({len(limits)}) and number of i | validation | error | combined-loader, limits, length-mismatch |
| Please call `iter(combined_loader)` first. | validation | error | combined-loader, iteration-order, runtime-state |
| All datasets are iterable-style datasets. | validation | error | combined-loader, iterable-dataset, not-implemented |
| The CombinedLoader has {len(stateful_loaders)} stateful load | validation | critical | combined-loader, checkpoint, state-mismatch, resume |
| `model` is required to be a `OptimizedModule`. Found a `{typ | validation | error | torch-compile, type-mismatch, unwrap |
| `model` is expected to be a compiled LightningModule. Found | validation | error | torch-compile, mixed-imports, type-mismatch |
| Unexpected error, the wrapped model should be a LightningMod | validation | error | torch-compile, unwrap, invariant-violation |
| `model` is required to be a compiled LightningModule. Found | validation | error | torch-compile, state-check |
| `model` must either be an instance of OptimizedModule or Lig | validation | error | torch-compile, type-mismatch |
| `model` must be a `LightningModule` or `torch._dynamo.Optimi | validation | error | torch-compile, mixed-imports, trainer, type-mismatch |
| Using a compiled model is incompatible with the current stra | validation | error | pytorch-lightning, torch-compile, strategy, distributed-training |
| `{dataloader_cls_name}` within local rank has zero length. P | validation | error | pytorch-lightning, dataloader, distributed-training, empty-dataset |
| The dataloader {dataloader} needs to subclass `torch.utils.d | validation | error | pytorch-lightning, dataloader, type-validation |
| Trying to inject custom `Sampler` into the `{dataloader_cls_ | exception | error | pytorch-lightning, dataloader, sampler, introspection |
| Trying to inject parameters into the `{dataloader_cls_name}` | exception | error | pytorch-lightning, dataloader, kwargs, signature-mismatch |
| Trying to inject a modified sampler into the batch sampler; | exception | error | pytorch-lightning, batch-sampler, sampler, distributed-training |
| Lightning can't inject a (distributed) sampler into your ba | exception | error | pytorch-lightning, batch-sampler, distributed-training, typeerror |
| Lightning can't inject a (distributed) sampler into your ba | exception | error | pytorch-lightning, batch-sampler, distributed-training, typeerror |
| Unable to find 'latest' file at {latest_path} | exception | error | pytorch-lightning, deepspeed, checkpoint, conversion |
| Directory '{ds_checkpoint_dir}' doesn't exist | exception | error | pytorch-lightning, deepspeed, checkpoint, file-not-found |
| Download model failed - {model_registry} | exception | critical | model-registry, download, lightning, checkpoint |
| `max_depth` can be -1, 0 or > 0, got {max_depth}. | exception | error | model-summary, validation, lightning, valueerror |
| There is no `frame` available while being required. | exception | error | hyperparameters, introspection, lightning, attributeerror |
| {attribute} is neither stored in the model namespace nor the | exception | error | attribute-lookup, batch-size, tuner, lightning |
| You have overridden the `LightningModule.backward` hook but | console | warning | deepspeed, backward-hook, precision, lightning |
| The PyTorch Profiler default schedule will be overridden as | console | warning | profiler, pytorch-profiler, schedule, lightning |
| You have overridden `{hook_name}` in both `LightningModule` | console | warning | datamodule, hook-conflict, data-loading, lightning |
| You have overridden `{hook_name}` in `LightningModule` but h | console | warning | datamodule, hook-conflict, data-loading, lightning |
| The dirpath has changed from {dirpath_from_ckpt!r} to {self. | console | warning | model-checkpoint, resume, dirpath, lightning |
| Skipping '{k}' parameter because it is not possible to safel | console | warning | hyperparameters, yaml-serialization, saving, lightning |
| LitLogger does not support `log_graph` | console | warning | logger, log-graph, litlogger, lightning |
| `training_step` returned `None`. If this was on purpose, ign | console | warning | training-step, loss, automatic-optimization, lightning |
| predict returned None if it was on purpose, ignore this warn | console | warning | predict-step, inference, lightning |
| Couldn't infer the batch indices fetched from your dataloade | console | warning | prediction, dataloader, batch-indices, lightning |
| train_dataloader yielded None. If this was on purpose, ignor | console | warning | training, dataloader, none-batch, lightning |
| When saving the DeepSpeed Stage 3 checkpoint, each worker wi | console | warning | deepspeed, zero-stage-3, checkpoint, sharding |
| Starting from v1.9.0, `tensorboardX` has been removed as a d | console | warning | logger, tensorboard, csv-fallback, lightning, version-change |
| You called `self.log({self.meta.name!r}, ...)` in your `{sel | console | warning | logging, dtype, metrics, lightning |
| It is recommended to use `self.log({result_metric.meta.name! | console | warning | pytorch-lightning, distributed, ddp, logging, sync-dist, metrics |
| Trying to infer the `batch_size` from an ambiguous collectio | console | warning | pytorch-lightning, batch-size, logging, data-loading |
| Redirecting import of {module}.{name} to {new_module}.{name} | console | info | pytorch-lightning, migration, pickle, checkpoint, version-upgrade |
| The total number of parameters detected may be inaccurate be | console | warning | pytorch-lightning, model-summary, lazy-layers, uninitialized-parameter, parameter-counting |