mirror of
https://github.com/PaddlePaddle/FastDeploy.git
synced 2026-04-24 01:29:57 +08:00
jackyYang6
988e0bc338
* feat(paddleformers): add dense text model fallback backend * docs(paddleformers): add user guide and fix code review issues * add fallback unit test * precommit format * fix pre-commit * fix: address code review feedback * docs: add PaddleFormers backend documentation (EN) and simplify installation --------- Co-authored-by: Jiang-Jia-Jun <163579578+Jiang-Jia-Jun@users.noreply.github.com> Co-authored-by: YuBaoku <49938469+EmmonsCurse@users.noreply.github.com>
72 lines
2.4 KiB
Python
72 lines
2.4 KiB
Python
"""
|
|
# Copyright (c) 2025 PaddlePaddle Authors. All Rights Reserved.
|
|
#
|
|
# Licensed under the Apache License, Version 2.0 (the "License");
|
|
# you may not use this file except in compliance with the License.
|
|
# You may obtain a copy of the License at
|
|
#
|
|
# http://www.apache.org/licenses/LICENSE-2.0
|
|
#
|
|
# Unless required by applicable law or agreed to in writing, software
|
|
# distributed under the License is distributed on an "AS IS" BASIS,
|
|
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
|
# See the License for the specific language governing permissions and
|
|
# limitations under the License.
|
|
"""
|
|
|
|
"""Causal LM Mixin for PaddleFormers models.
|
|
This mixin provides lm_head and compute_logits functionality.
|
|
The forward() method is implemented in PaddleFormersModelBase.
|
|
"""
|
|
|
|
from typing import TYPE_CHECKING
|
|
|
|
import paddle
|
|
|
|
from fastdeploy.model_executor.layers.lm_head import ParallelLMHead
|
|
|
|
if TYPE_CHECKING:
|
|
from fastdeploy.config import FDConfig
|
|
|
|
|
|
class CausalLMMixin:
|
|
"""Mixin class that provides causal LM functionality for PaddleFormers models.
|
|
|
|
This mixin only handles:
|
|
- lm_head initialization
|
|
- compute_logits (hidden_states -> logits)
|
|
|
|
The forward() method is inherited from PaddleFormersModelBase which computes
|
|
input_ids -> hidden_states.
|
|
|
|
This is a private mixin class and should NOT be instantiated directly.
|
|
Use PaddleFormersForCausalLM instead.
|
|
"""
|
|
|
|
def __init__(self, fd_config: "FDConfig", **kwargs):
|
|
|
|
super().__init__(fd_config, **kwargs)
|
|
|
|
self.ori_vocab_size = fd_config.model_config.ori_vocab_size
|
|
self.tie_word_embeddings = fd_config.model_config.tie_word_embeddings
|
|
with_bias = getattr(self.text_config, "use_bias", False) or getattr(self.text_config, "bias", False)
|
|
|
|
self.lm_head = ParallelLMHead(
|
|
fd_config=fd_config,
|
|
embedding_dim=self.text_config.hidden_size,
|
|
num_embeddings=self.text_config.vocab_size,
|
|
prefix="lm_head",
|
|
with_bias=with_bias,
|
|
)
|
|
|
|
def compute_logits(self, hidden_state, **kwargs):
|
|
"""Compute logits from hidden states using lm_head."""
|
|
logits = self.lm_head(hidden_state)
|
|
logits = logits.astype(paddle.float32)
|
|
logits[:, self.ori_vocab_size :] = -float("inf")
|
|
|
|
return logits
|
|
|
|
def set_state_dict(self, state_dict):
|
|
self.load_weights(state_dict.items())
|