| 1 | # SPDX-License-Identifier: Apache-2.0 |
| 2 | # SPDX-FileCopyrightText: Copyright contributors to the vLLM project |
| 3 | """Config definitions for ExtractHiddenStatesModel, to be used with |
| 4 | the extract_hidden_states spec decoding method.""" |
| 5 | |
| 6 | import os |
| 7 | |
| 8 | from transformers import PreTrainedConfig |
| 9 | |
| 10 | from vllm.transformers_utils.utils import without_trust_remote_code |
| 11 | |
| 12 | |
| 13 | class ExtractHiddenStatesConfig(PreTrainedConfig): |
| 14 | model_type = "extract_hidden_states" |
| 15 | |
| 16 | def __init__( |
| 17 | self, |
| 18 | model: PreTrainedConfig | dict | None = None, |
| 19 | method: str | None = "extract_hidden_states", |
| 20 | **kwargs, |
| 21 | ): |
| 22 | assert method == "extract_hidden_states" |
| 23 | |
| 24 | if isinstance(model, dict): |
| 25 | model_dict = model |
| 26 | source_text_config = None |
| 27 | elif isinstance(model, PreTrainedConfig): |
| 28 | model_dict = model.to_dict() |
| 29 | text_config = model.get_text_config() |
| 30 | source_text_config = text_config if text_config is not model else None |
| 31 | else: |
| 32 | model_dict = {} |
| 33 | source_text_config = None |
| 34 | |
| 35 | # Combine: model_dict first, then kwargs override |
| 36 | combined = {**model_dict, **kwargs} |
| 37 | # Remove architectures from the base, we'll set it explicitly |
| 38 | combined = {k: v for k, v in combined.items() if k != "architectures"} |
| 39 | |
| 40 | combined["architectures"] = ["ExtractHiddenStatesModel"] |
| 41 | |
| 42 | # to_dict() and kwargs both flatten text_config to a plain dict; |
| 43 | # downstream get_hf_text_config() needs it as a PreTrainedConfig |
| 44 | # for attribute access. Re-insert the original object. |
| 45 | if source_text_config is not None: |
| 46 | combined["text_config"] = source_text_config |
| 47 | |
| 48 | super().__init__(**combined) |
| 49 | |
| 50 | @classmethod |
| 51 | def from_pretrained( |
| 52 | cls, |
| 53 | pretrained_model_name_or_path: str | os.PathLike, |
| 54 | **kwargs, |
| 55 | ) -> "ExtractHiddenStatesConfig": |
| 56 | config_dict, kwargs = cls.get_config_dict( |
| 57 | pretrained_model_name_or_path, **without_trust_remote_code(kwargs) |
| 58 | ) |
| 59 | return cls.from_dict(config_dict, **kwargs) |
| 60 | |
| 61 | def to_json_string(self, use_diff: bool = True) -> str: |
| 62 | # we override use_diff to False as initializing |
| 63 | # ExtractHiddenStatesConfig with default arguments is not supported |
| 64 | del use_diff |
| 65 | return super().to_json_string(use_diff=False) |