df1ee8070d
### What this PR does / why we need it?
Implement a unified parallelized speculative decoding in VLLM
Ascend，which can simultaneously support parallel speculative inference
schemes such as Pard, P-Eagle, etc. refer to
https://github.com/vllm-project/vllm-ascend/pull/6565 and
https://github.com/vllm-project/vllm-ascend/pull/4078
### How was this patch tested?
run with parallel drafting script:
export target=/model/Llama-3.1-8B-Instruct
export draft=/model/PARD-Llama-3.2-1B
export CUDA_VISIBLE_DEVICES=6
export ASCEND_RT_VISIBLE_DEVICES=6
vllm serve $target \
--tensor-parallel-size 1 \
--max-model-len 4096 \
--no-enable-prefix-caching \
--port 8811 \
--speculative-config '{"model": "/model/PARD-Llama-3.2-1B", "method":
"draft_model", "num_speculative_tokens": 8, "parallel_drafting": true}'
base script:
export target=/model/Llama-3.1-8B-Instruct
export draft=/model/PARD-Llama-3.2-1B
export CUDA_VISIBLE_DEVICES=6
export ASCEND_RT_VISIBLE_DEVICES=6
vllm serve $target \
--tensor-parallel-size 1 \
--max-model-len 4096 \
--no-enable-prefix-caching \
--port 8811
benchmark script:
MAX_CONCURRENCY=1
NUM_PROMPTS=80
vllm bench serve --port 8811 \
--temperature 0 \
--model /model/Llama-3.1-8B-Instruct \
--backend openai-chat \
--endpoint /v1/chat/completions \
--dataset-name hf \
--dataset-path philschmid/mt-bench \
--num-prompts ${NUM_PROMPTS} \
--max-concurrency ${MAX_CONCURRENCY} \
--seed 1234
test results :
base(without spec decode): TTFT 79.46ms TPOT 26.99ms
output_tokens_throughput 36.75 tok/s
this pr(with parallel drafting): TTFT 72.24ms TPOT 13.45ms
output_tokens_throughput 72.98 tok/s
per-position acceptance(from position 0 to 7):
79.48%、56.93%、40%、27.90%、19.79%、14.25%、10.57%、7.61%.
----------------------------------------------------------------------
run on qwen3 model script ：
export target=/model/Qwen3-1.7B
export draft=/model/PARD-Qwen3-0.6B
export CUDA_VISIBLE_DEVICES=1
export ASCEND_RT_VISIBLE_DEVICES=1
vllm serve $target \
--tensor-parallel-size 1 \
--max-model-len 4096 \
--no-enable-prefix-caching \
--port 8811 \
--speculative-config '{"model": "/model/PARD-Qwen3-0.6B", "method":
"draft_model", "num_speculative_tokens": 8, "parallel_drafting": true}'
cc @NickJudyHvv
- vLLM version: v0.15.0
- vLLM main:
9562912cea
---------
Signed-off-by: 01267596 <xiongkai123@cmbchina.com>
Signed-off-by: kx <1670186653@qq.com>
Signed-off-by: HF-001 <1670186653@qq.com>
Co-authored-by: 01267596 <xiongkai123@cmbchina.com>
40 lines
1.7 KiB
Python
40 lines
1.7 KiB
Python
#
|
|
# Copyright (c) 2025 Huawei Technologies Co., Ltd. All Rights Reserved.
|
|
# Copyright 2023 The vLLM team.
|
|
#
|
|
# 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.
|
|
# This file is a part of the vllm-ascend project.
|
|
# Adapted from vllm-project/vllm/vllm/worker/gpu_model_runner.py
|
|
#
|
|
|
|
from vllm_ascend.spec_decode.draft_proposer import AscendDraftModelProposer
|
|
from vllm_ascend.spec_decode.eagle_proposer import AscendEagleProposer
|
|
from vllm_ascend.spec_decode.medusa_proposer import AscendMedusaProposer
|
|
from vllm_ascend.spec_decode.ngram_proposer import AscendNgramProposer
|
|
from vllm_ascend.spec_decode.suffix_proposer import AscendSuffixDecodingProposer
|
|
|
|
|
|
def get_spec_decode_method(method, vllm_config, device, runner):
|
|
if method == "ngram":
|
|
return AscendNgramProposer(vllm_config, runner)
|
|
elif method == "suffix":
|
|
return AscendSuffixDecodingProposer(vllm_config, runner)
|
|
elif method == "medusa":
|
|
return AscendMedusaProposer(vllm_config, device)
|
|
elif method in ("eagle", "eagle3", "mtp"):
|
|
return AscendEagleProposer(vllm_config, device, runner)
|
|
elif method == "draft_model":
|
|
return AscendDraftModelProposer(vllm_config, device, runner)
|
|
else:
|
|
raise ValueError(f"Unknown speculative decoding method: {method}")
|