This class includes only the parameters that are specific to SFT training. For a full list of training arguments,
please refer to the [`~transformers.TrainingArguments`] documentation. Note that default values in this class may
differ from those in [`~transformers.TrainingArguments`].
Using [`~transformers.HfArgumentParser`] we can turn this class into
[argparse](https://docs.python.org/3/library/argparse#module-argparse) arguments that can be specified on the
command line.
Parameters:
> Parameters that control the model
model_init_kwargs (`dict[str, Any]` or `None`, *optional*, defaults to `None`):
Keyword arguments for [`~transformers.AutoModelForCausalLM.from_pretrained`], used when the `model`
argument of the [`SFTTrainer`] is provided as a string.
chat_template_path (`str` or `None`, *optional*, defaults to `None`):
If specified, sets the model's chat template. This can either be the path to a tokenizer (local directory
or Hugging Face Hub model) or a direct path to a Jinja template file. When using a Jinja file, you must
ensure that any special tokens referenced in the template are added to the tokenizer and that the model's
embedding layer is resized accordingly.
> Parameters that control the data preprocessing
dataset_text_field (`str`, *optional*, defaults to `"text"`):
Name of the column that contains text data in the dataset.
dataset_kwargs (`dict[str, Any]` or `None`, *optional*, defaults to `None`):
Dictionary of optional keyword arguments for the dataset preparation. The only supported key is
`skip_prepare_dataset`.
dataset_num_proc (`int` or `None`, *optional*, defaults to `None`):
Number of processes to use for processing the dataset.
eos_token (`str` or `None`, *optional*, defaults to `None`):
Token used to indicate the end of a turn or sequence. If `None`, it defaults to
`processing_class.eos_token`.
pad_token (`int` or `None`, *optional*, defaults to `None`):
Token used for padding. If `None`, it defaults to `processing_class.pad_token`, or if that is also `None`,
it falls back to `processing_class.eos_token`.
max_length (`int` or `None`, *optional*, defaults to `1024`):
Maximum length of the tokenized sequence. Sequences longer than `max_length` are truncated from the right.
If `None`, no truncation is applied. When packing is enabled, this value sets the sequence length.
packing (`bool`, *optional*, defaults to `False`):
Whether to group multiple sequences into fixed-length blocks to improve computational efficiency and reduce
padding. Uses `max_length` to define sequence length.
packing_strategy (`str`, *optional*, defaults to `"bfd"`):
Strategy for packing sequences. Can be either `"bfd"` (best-fit decreasing, default), or `"wrapped"`.
padding_free (`bool`, *optional*, defaults to `False`):
Whether to perform forward passes without padding by flattening all sequences in the batch into a single
continuous sequence. This reduces memory usage by eliminating padding overhead. Currently, this is only
supported with the FlashAttention 2 or 3, which can efficiently handle the flattened batch structure. When
packing is enabled with strategy `"bfd"`, padding-free is enabled, regardless of the value of this
parameter.
pad_to_multiple_of (`int` or `None`, *optional*, defaults to `None`):
If set, the sequences will be padded to a multiple of this value.
eval_packing (`bool` or `None`, *optional*, defaults to `None`):
Whether to pack the eval dataset. If `None`, uses the same value as `packing`.
> Parameters that control the training
completion_only_loss (`bool` or `None`, *optional*, defaults to `None`):
Whether to compute loss only on the completion part of the sequence. If set to `True`, loss is computed
only on the completion, which is supported only for [prompt-completion](#prompt-completion) datasets. If
`False`, loss is computed on the entire sequence. If `None` (default), the behavior depends on the dataset:
loss is computed on the completion for [prompt-completion](#prompt-completion) datasets, and on the full
sequence for [language modeling](#language-modeling) datasets.
assistant_only_loss (`bool`, *optional*, defaults to `False`):
Whether to compute loss only on the assistant part of the sequence. If set to `True`, loss is computed
only on the assistant responses, which is supported only for [conversational](#conversational) datasets. If `False`,
loss is computed on the entire sequence.
activation_offloading (`bool`, *optional*, defaults to `False`):
Whether to offload the activations to the CPU.
NÚhelpzvLLM SamplingParams)ÚdefaultÚmetadataÚvllm_sampling_paramsrBz8Chunk size to reduce memory usage. -1 is most efficient.Úunsloth_num_chunksz'Maximum sequence length to truncate to.Úmax_seq_lengthFÚnorCéréúç-Cëâ6
model_pathÚreturncCsv|jpi}|d¡}t|tjƒs|dks|durnt|tƒr(tt|ƒ}||d<ntd|›d�ƒ‚tj |fi|¤Ž}|S)z0Creates a model from a path or model identifier.Útorch_dtyperyNzˆInvalid `torch_dtype` passed to `SFTConfig`. Expected either 'auto' or a string representing a `torch.dtype` (e.g., 'float32'), but got Ú.)
�}nt||ƒ}|jr�t|ddƒr�|s�t|ƒ|S)z#Prepares a model for PEFT training.z9To use PeftModel, you need to install the `peft` library.Úis_loaded_in_4bitFÚis_loaded_in_8bitÚ
z&_UnslothSFTTrainer._prepare_peft_modelcCs"|j|jpidœ}t|fi|¤ŽS)z-Prepares a quantized model for kbit training.)Úuse_gradient_checkpointingrò)rñròr2)r%r4r5Úprepare_model_kwargsr]r]r^r�ïsþz3_UnslothSFTTrainer._prepare_model_for_kbit_trainingcCsN|jpi}d|vp|d}|r%t|dƒr|¡|Sdd„}|¡|¡|S)z-Enables gradient checkpointing for the model.Ú
i|¤Ž}| �r�t|dƒ�s�|ˆd d�}||_|�rÆztWntd%ƒ|YSˆdk�rtd&ƒ‚|�r¸d'|›d(�|d#<t||¡ˆt|d)d*ƒ|ƒ}|S)+NÚ tokenizerrrrfÚmax_seqz1Unsloth: max_seq_length is 0! Please specify one!rr�FTÚ input_idsÚattention_maskr:Úlabelsr/z Unsloth: z does not have .pad!)Úmlmz-Unsloth: You must specify a `formatting_func`zIUnsloth: The `formatting_func` should return a list of processed strings.rgrzÚ bos_tokenzHUnsloth: We found double BOS tokens - we shall remove one automatically.cs"ˆˆs|ˆnˆ|ƒˆˆdˆd�S)NF)Ú
batch_sizezUnsloth: Tokenizing ["z"]ÚdescÚbatchedzPUnsloth: Hugging Face's packing is currently buggy - we're disabling it for now!z:When packing is enabled, `max_seq_length` can't be `None`.zUnsloth: Packing z datasetrrƒ)rUÚConstantLengthDatasetrrtr¥rƒÚRuntimeErrorÚsetrlrmÚkeysrRÚtransformersr;rr(r6rpr^Ú
