f6149f3894
This is the follow-up PR to PR #3189, which continues to refactor sfa
into mla and finally remove deepseek_v3_2.py. This is the last PR of
deepseek modeling refactoring. After this, all deepseek-related model
codes are removed from vllm_ascend.
FurtherMore, after this PR deepseek v3.2 can run chunk-prefill with
correct accuracy.
- vLLM version: v0.11.0rc3
- vLLM main:
83f478bb19
---------
Signed-off-by: whx-sjtu <2952154980@qq.com>
109 lines
4.0 KiB
Python
109 lines
4.0 KiB
Python
#
|
|
# Copyright (c) 2025 Huawei Technologies Co., Ltd. All Rights Reserved.
|
|
# This file is a part of the vllm-ascend project.
|
|
#
|
|
# 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.
|
|
#
|
|
from itertools import islice
|
|
from typing import Optional, Union
|
|
|
|
import torch
|
|
import vllm.model_executor.models.deepseek_v2
|
|
from torch import nn
|
|
from vllm.compilation.decorators import support_torch_compile
|
|
from vllm.config import VllmConfig
|
|
from vllm.distributed import get_pp_group
|
|
from vllm.model_executor.layers.layernorm import RMSNorm
|
|
from vllm.model_executor.layers.vocab_parallel_embedding import \
|
|
VocabParallelEmbedding
|
|
from vllm.model_executor.models.deepseek_v2 import DeepseekV2DecoderLayer
|
|
from vllm.model_executor.models.utils import (
|
|
PPMissingLayer, make_empty_intermediate_tensors_factory, make_layers)
|
|
from vllm.sequence import IntermediateTensors
|
|
|
|
|
|
@support_torch_compile
|
|
class DeepseekV2Model(nn.Module):
|
|
|
|
fall_back_to_pt_during_load = False
|
|
|
|
def __init__(self, *, vllm_config: VllmConfig, prefix: str = ""):
|
|
super().__init__()
|
|
|
|
config = vllm_config.model_config.hf_config
|
|
quant_config = vllm_config.quant_config
|
|
self.config = config
|
|
|
|
self.vocab_size = config.vocab_size
|
|
self.is_v32 = hasattr(config, "index_topk")
|
|
topk_indices_buffer = None
|
|
|
|
if get_pp_group().is_first_rank:
|
|
self.embed_tokens = VocabParallelEmbedding(
|
|
config.vocab_size,
|
|
config.hidden_size,
|
|
quant_config=quant_config,
|
|
prefix=f"{prefix}.embed_tokens")
|
|
else:
|
|
self.embed_tokens = PPMissingLayer()
|
|
|
|
self.start_layer, self.end_layer, self.layers = make_layers(
|
|
config.num_hidden_layers,
|
|
lambda prefix: DeepseekV2DecoderLayer(vllm_config, prefix,
|
|
topk_indices_buffer),
|
|
prefix=f"{prefix}.layers")
|
|
|
|
if get_pp_group().is_last_rank:
|
|
self.norm = RMSNorm(config.hidden_size, eps=config.rms_norm_eps)
|
|
else:
|
|
self.norm = PPMissingLayer()
|
|
self.make_empty_intermediate_tensors = (
|
|
make_empty_intermediate_tensors_factory(
|
|
["hidden_states", "residual"], config.hidden_size))
|
|
|
|
def get_input_embeddings(self, input_ids: torch.Tensor) -> torch.Tensor:
|
|
return self.embed_tokens(input_ids)
|
|
|
|
def forward(
|
|
self,
|
|
input_ids: torch.Tensor,
|
|
positions: torch.Tensor,
|
|
intermediate_tensors: Optional[IntermediateTensors],
|
|
inputs_embeds: Optional[torch.Tensor] = None,
|
|
) -> Union[torch.Tensor, IntermediateTensors]:
|
|
if get_pp_group().is_first_rank:
|
|
if inputs_embeds is not None:
|
|
hidden_states = inputs_embeds
|
|
else:
|
|
hidden_states = self.get_input_embeddings(input_ids)
|
|
residual = None
|
|
else:
|
|
assert intermediate_tensors is not None
|
|
hidden_states = intermediate_tensors["hidden_states"]
|
|
residual = intermediate_tensors["residual"]
|
|
|
|
for layer in islice(self.layers, self.start_layer, self.end_layer):
|
|
hidden_states, residual = layer(positions, hidden_states, residual)
|
|
|
|
if not get_pp_group().is_last_rank:
|
|
return IntermediateTensors({
|
|
"hidden_states": hidden_states,
|
|
"residual": residual
|
|
})
|
|
|
|
hidden_states, _ = self.norm(hidden_states, residual)
|
|
return hidden_states
|
|
|
|
|
|
vllm.model_executor.models.deepseek_v2.DeepseekV2Model = DeepseekV2Model
|