peft/pr_3202/en/package_reference/auto_class.md

AutoPeftModels

The AutoPeftModel classes loads the appropriate PEFT model for the task type by automatically inferring it from the configuration file. They are designed to quickly and easily load a PEFT model in a single line of code without having to worry about which exact model class you need or manually loading a PeftConfig.

AutoPeftModel[[peft.AutoPeftModel]]

peft.AutoPeftModel[[peft.AutoPeftModel]]

Source

from_pretrainedpeft.AutoPeftModel.from_pretrainedhttps://github.com/huggingface/peft/blob/vr_3202/src/peft/auto.py#L88[{"name": "pretrained_model_name_or_path", "val": ""}, {"name": "adapter_name", "val": ": str = 'default'"}, {"name": "is_trainable", "val": ": bool = False"}, {"name": "config", "val": ": Optional[PeftConfig] = None"}, {"name": "revision", "val": ": Optional[str] = None"}, {"name": "import_allowlist", "val": ": Optional[list[str]] = None"}, {"name": "**kwargs", "val": ""}]- import_allowlist (list[str], optional, defaults to {get_default_import_allowlist()}) -- AutoPeftModel will attempt to instantiate the base model that is configured in the adapter config. Since this operation needs to potentially import other packages, this allowlist is a safe-guard to prevent importing malicious packages. You may need to specify your package's import name here if it is not in the defaults.0

A wrapper around all the preprocessing steps a user needs to perform in order to load a PEFT model. The kwargs are passed along to PeftConfig that automatically takes care of filtering the kwargs of the Hub methods and the config object init.

The parameters are equivalent to the ones of PeftModel.from_pretrained(). Differences are documented below.

Parameters:

import_allowlist (list[str], optional, defaults to {get_default_import_allowlist()}) : AutoPeftModel will attempt to instantiate the base model that is configured in the adapter config. Since this operation needs to potentially import other packages, this allowlist is a safe-guard to prevent importing malicious packages. You may need to specify your package's import name here if it is not in the defaults.

AutoPeftModelForCausalLM[[peft.AutoPeftModelForCausalLM]]

peft.AutoPeftModelForCausalLM[[peft.AutoPeftModelForCausalLM]]

Source

AutoPeftModelForSeq2SeqLM[[peft.AutoPeftModelForSeq2SeqLM]]

peft.AutoPeftModelForSeq2SeqLM[[peft.AutoPeftModelForSeq2SeqLM]]

Source

AutoPeftModelForSequenceClassification[[peft.AutoPeftModelForSequenceClassification]]

peft.AutoPeftModelForSequenceClassification[[peft.AutoPeftModelForSequenceClassification]]

Source

AutoPeftModelForTokenClassification[[peft.AutoPeftModelForTokenClassification]]

peft.AutoPeftModelForTokenClassification[[peft.AutoPeftModelForTokenClassification]]

Source

AutoPeftModelForQuestionAnswering[[peft.AutoPeftModelForQuestionAnswering]]

peft.AutoPeftModelForQuestionAnswering[[peft.AutoPeftModelForQuestionAnswering]]

Source

AutoPeftModelForFeatureExtraction[[peft.AutoPeftModelForFeatureExtraction]]

peft.AutoPeftModelForFeatureExtraction[[peft.AutoPeftModelForFeatureExtraction]]

Source