EngineX/xc-llm-ascend
3
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

[Spec Decode]clean up spec decode interface (#6947)

Browse Source 
This pull request refactors the speculative decoding proposer interface
to align with upstream vLLM, removing the local `Proposer` interface and
renaming methods to `propose`.

This is the first step. In the future we should remove the class
register and just add few Ascend specified method once the arch in vLLM
is ready.

- vLLM version: v0.16.0
- vLLM main:
15d76f74e2

Signed-off-by: wangxiyuan <wangxiyuan1007@gmail.com>
This commit is contained in:
wangxiyuan
2026-03-05 14:30:10 +08:00
committed by GitHub
parent 2bd9c35788
commit 13777bf3f0
11 changed files with 194 additions and 315 deletions
Download Patch File Download Diff File Expand all files Collapse all files

24
vllm_ascend/spec_decode/__init__.py
View File

@@ -16,23 +16,23 @@
# 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.eagle_proposer import EagleProposer
from vllm_ascend.spec_decode.medusa_proposer import MedusaProposer
from vllm_ascend.spec_decode.mtp_proposer import MtpProposer
from vllm_ascend.spec_decode.ngram_proposer import NgramProposer
from vllm_ascend.spec_decode.suffix_proposer import SuffixDecodingProposer
from vllm_ascend.spec_decode.eagle_proposer import AscendEagleProposer
from vllm_ascend.spec_decode.medusa_proposer import AscendMedusaProposer
from vllm_ascend.spec_decode.mtp_proposer import AscendMtpProposer
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 NgramProposer(vllm_config, device, runner)
elif method in ("eagle", "eagle3"):
return EagleProposer(vllm_config, device, runner)
elif method == "mtp":
return MtpProposer(vllm_config, device, runner)
return AscendNgramProposer(vllm_config, runner)
elif method == "suffix":
return SuffixDecodingProposer(vllm_config, device, runner)
return AscendSuffixDecodingProposer(vllm_config, runner)
elif method == "medusa":
return MedusaProposer(vllm_config, device, runner)
return AscendMedusaProposer(vllm_config, device)
elif method in ("eagle", "eagle3"):
return AscendEagleProposer(vllm_config, device, runner)
elif method == "mtp":
return AscendMtpProposer(vllm_config, device, runner)
else:
raise ValueError(f"Unknown speculative decoding method: {method}")