ErrLookup › hiyouga/LlamaFactory
hiyouga/LlamaFactory
Unified Efficient Fine-Tuning of 100+ LLMs & VLMs (ACL 2024) · Python · 259 source files
Analyzed at f28afaf635 on 2026-08-14. 386 documented errors.
| Code / Message | Type | Severity | Tags |
|---|---|---|---|
| Plugin is not registered under . | validation | error | plugin-registry, llamafactory, v1-plugins, import, config |
| Unknown params for . : . Expected | validation | error | config, validation, llamafactory, v1-plugins, yaml |
| prompt is not a token-prefix of the full sequence; the chat… | validation | error | chat-template, tokenization, prefix-stability, training-data, labels |
| SGLang server initialization failed | exception | error | sglang, server, subprocess, inference, startup |
| Merged position_ids shape mismatch: got | exception | error | collator, mrope, multimodal, tensor-shape, training |
| Invalid API key. | http | error | megatron, model-support, version, config |
| does not support gradient checkpointing. | exception | error | gradient-checkpointing, vram, custom-model, training-config |
| Module not found in hidden modules | validation | error | peft, freeze-tuning, module-names, configuration |
| Qwen2VL requires 3D position ids for mrope. | exception | error | dependency, qwen3-5, packing, flash-linear-attention, installation |
| Logits (batchsize x seqlen) and labels must have the same… | exception | critical | dpo, logits, shape-mismatch, preference-training |
| Failed to export model: weight conversion reversal is not… | exception | error | export, transformers-5, version-incompatibility, mistral |
| Only 4-bit quantized model can use fsdp+qlora or auto… | validation | error | quantization, bitsandbytes, qlora, export, bug |
| `extra_config` file not found | validation | error | megatron, file-not-found, config-validation |
| Input must be string, set[str] or dict[str, str], got | exception | error | templates, data-format, custom-template |
| `recompute_method` must be 'uniform' or 'block'. | validation | error | megatron, activation-checkpointing, config-validation |
| Not allowed | http | error | megatron-bridge, chat-template, jinja, dataset |
| YAML config must be a dictionary mapping tokens to… | validation | error | hparams, tokenizer, special-tokens, yaml, config-validation |
| Cannot process the logits. | exception | error | sft, evaluation, logits, metrics |
| Processor did not emit | exception | error | multimodal, processor, dummy-data, compatibility, transformers-version |
| These `kt_config` values are derived from LLaMA-Factory… | validation | error | ktransformers, config-conflict, derived-fields |
| special-token escape failed: the tokenizer normalized away… | validation | error | tokenizer, special-tokens, data-cleaning, prompt-injection, rendering |
| The `master_addr` ( ) is not in Ray cluster or not alive | exception | error | ray, distributed-training, configuration, network |
| The length of packed example should be identical to the… | exception | critical | data, packing, sft, config |
| Stage does not supported | exception | error | kto, dataset, data-quality, collator |
| Cannot create new adapter upon a quantized model. | validation | error | config, qlora, adapter, validation, llamafactory |
| torch.accelerator is not available, please upgrade torch to… | exception | error | torch, version-mismatch, v1, environment |
| Cannot use PiSSA for current training stage. | validation | error | config, pissa, lora, ppo, kto, dpo |
| Unknown attention type | exception | error | flash-attention, config, enum, attention |
| batching_strategy= does not support multimodal data; use… | exception | error | batching, multimodal, packing, not-implemented, training |
| Invalid length | http | error | megatron-bridge, model-support, multimodal, config |
| `kt_cpu_activation: recompute` requires GPU gradient… | validation | error | ktransformers, gradient-checkpointing, memory |
| The installed kt-kernel does not provide the activation… | exception | error | kt-kernel, dependency, gradient-checkpointing, environment |
| `reward_model_type` cannot be lora for Freeze/Full PPO… | validation | error | config, ppo, lora, reward-model |
| Failed to load tokenizer. | exception | critical | tokenizer, huggingface-hub, network, auth, offline |
| Please upgrade `transformers` to 4.34.0 | exception | error | model, internvl, multimodal, checkpoint-format, config |
| `resume_from_checkpoint` will be supported in the future… | exception | error | ppo, reinforcement-learning, checkpoint, resume |
| Cannot use GaLore, APOLLO or BAdam together. | validation | error | config, galore, apollo, badam, optimizer |
| requires 3D position ids for mrope. | exception | error | mrope, collator, qwen-vl, multimodal, training |
| Streaming mode should have an integer val size. | validation | error | config, streaming, data, validation-split |
| DeepSpeed ZeRO-3 model-loading bootstrap failed… | exception | critical | deepspeed, environment, distributed, zero3, version-conflict |
| Invalid JSON format in function message | exception | error | json, tool-calls, data-format, sharegpt |
| kernel_config.include_kernels must be 'auto' or a… | validation | error | configuration, type-error, kernel-plugin, linear-attention |
| `virtual_pipeline_model_parallel_size` must be >= 1 when… | validation | error | megatron, parallelism, config-validation, hparams |
| placeholder count ( ) != number of blocks ( ); media must… | validation | error | multimodal, placeholders, data-format, rendering, vision |
| `moe_token_dispatcher_type` must be 'allgather'… | validation | error | megatron, moe, config-validation |
| Unsupported torch version detected: torch 2.9.x with… | exception | error | torch, compatibility, conv3d, multimodal, version-check |
| Current model does not support freeze tuning. | validation | error | peft, freeze-tuning, model-config, configuration |
| Quantized model only accepts a single adapter. Merge them… | validation | error | config, qlora, adapter, merging, validation |
| Current model does not support resizing embedding layers. | exception | error | embedding, tokenizer, lm-head, resize |
| Unexpected missing keys when loading checkpoint model… | exception | critical | checkpoint, resume, state-dict, model-mismatch |
| config must be a mapping or , got . | validation | error | plugin, config, type-error, validation |
| Total Megatron Bridge parallel size | validation | error | megatron, distributed, parallelism, config, validation |
| `sequence_parallel` requires `tensor_model_parallel_size` >… | validation | error | megatron, sequence-parallel, config-validation, hparams |
| Unknown optim: . | exception | error | galore, optimizer, sft, config |
| Current template does not support `train_on_prompt`. | exception | error | template, config, train-on-prompt |
| SGLang server error | exception | error | sglang, http, inference, api-error |
| Some specified arguments are not used by the… | exception | error | v1, configuration, cli, argument-parsing |
| `num_layers` should be divisible by `num_expand` . | exception | error | version, transformers, dpo, dependency |
| Checkpoint directory does not exist | validation | error | checkpoint, resume, filesystem, config |
| Cannot use LoRA with GaLore, APOLLO or BAdam together. | validation | error | config, lora, galore, apollo, badam, optimizer |
| Invalid or inaccessible file path. | http | error | security, filesystem, permissions, http-400, lfi |
| DPO training requires pair-format samples containing… | validation | error | data, dataset, dpo, preference-data, validation |
| FSDPTurbo expert parallelism requires an initialized… | exception | error | fsdpturbo, expert-parallel, device-mesh, initialization-order |
| KTransformers uses LLaMA-Factory's… | validation | error | ktransformers, gradient-checkpointing, config, llamafactory |
| `use_llama_pro` is only valid for Freeze or LoRA training. | validation | error | config, llama-pro, finetuning-type |
| Cannot resize embedding layers of a quantized model. | exception | error | quantization, tokenizer, embedding, resize |
| Module not found in non-hidden modules | validation | error | peft, freeze-tuning, module-names, configuration |
| `recompute_granularity` must be 'full' or 'selective'. | validation | error | megatron, activation-checkpointing, config-validation |
| Dict is not supported. | exception | error | template, jinja, export |
| Processor was not found, please check and update your model… | exception | error | multimodal, processor, model-files, config |
| Cannot find valid samples, check `data/README.md` for the… | exception | error | dataset, data-format, empty-dataset, preprocessing |
| Input must be string, set[str] or dict[str, str], got | exception | error | template, custom-template, data |
| MOSS-VL encountered nested video token blocks after… | exception | error | multimodal, moss-vl, video, truncation, cutoff-len |
| Put KTransformers settings in the LLaMA-Factory training… | validation | error | ktransformers, accelerate, config-conflict |
| Unexpected precision | validation | error | dtype, precision, validation, config |
| No FSDPTurbo EP spec is registered for model_type= | validation | error | distributed, fsdp, expert-parallelism, moe, config |
| hf_model_path=' ' does not exist locally and… | validation | error | fsdp2, model-loading, huggingface-hub, environment |
| `dpo_label_smoothing` is only valid for sigmoid loss… | validation | error | config, dpo, preference-learning, hyperparameters |
| KTransformers cannot be combined with Unsloth checkpoint… | validation | error | ktransformers, unsloth, incompatible-backends |
| Expect input is a list of images, but got | exception | error | multimodal, image, type-validation, data-preprocessing |
| FSDPTurbo parallel state must be initialized before… | exception | error | fsdpturbo, gradient-clipping, distributed, initialization-order |
| Quantization dataset is necessary for exporting. | validation | error | export, quantization, hparams, required-field |
| RM training requires pair data with token_type_ids. Ensure… | validation | error | reward-model, data, batch, token-type-ids |
| The model does not have a submodule named | exception | error | hyper-parallel, context-parallel, dataset, streaming |
| Currently lora stage does not support loading model by meta. | validation | error | lora, peft, model-loading, meta-device, v1 |
| Invalid JSON format in tool description | exception | error | json, tools, data-format, sharegpt |
| Unknown identifier | exception | error | tools, json, ast, glm4-moe |
| Only supports u/a/u/a/u... | http | error | dependency, megatron-bridge, installation, nemo |
| `padding_free` requires `flash_attn: flash_attention_2`. | exception | error | v1, flash-attention, batching, configuration |
| `loraplus_lr_ratio` is only valid for LoRA training. | validation | error | config, lora, loraplus, optimizer |
| The number of videos does not match the number of | exception | error | multimodal, video, dataset-validation, data-preprocessing |
| Cannot find satisfying example, considering decrease… | exception | error | gptq, export, quantization, calibration, dataset |
| Please specify peft_config to merge and export model. | validation | error | peft, export, merge, configuration |
| Image processor was not found, please check and update your… | exception | error | multimodal, image-processor, model-files, transformers |
| Unsupported model type | exception | error | dataset, config, hyper-parallel, training |
| Layer-wise BAdam only supports DeepSpeed ZeRO-3 training. | validation | error | badam, optimizer, deepspeed, zero-3, distributed, incompatible-flags |
| Omni models are not supported for packed sequences for now. | exception | error | packing, omni, collator, training, audio |
| The number of devices in the Ray cluster | exception | error | ray, distributed, gpu, resource-planning |
| Invalid tools | http | error | api, tools, function-calling, validation, http-400 |
| `kt_cpu_activation` must be `retain` or `recompute`. | validation | error | ktransformers, cpu-offload, config-validation |