biondizzle/vllm
1
0
Fork 0
Code Issues Pull Requests Actions 2 Packages Projects Releases Wiki Activity

[V1] Support Deepseek MTP (#18435)

Browse Source 
Signed-off-by: Rui Qiao <ruisearch42@gmail.com>
Signed-off-by: YaoJiayi <120040070@link.cuhk.edu.cn>
Co-authored-by: Rui Qiao <ruisearch42@gmail.com>
This commit is contained in:
Jiayi Yao
2025-05-23 12:26:28 -05:00
committed by GitHub
parent 371f7e4ca2
commit 2628a69e35
6 changed files with 120 additions and 66 deletions
Download Patch File Download Diff File Expand all files Collapse all files

13
vllm/config.py
View File

@@ -2255,7 +2255,7 @@ class DeviceConfig:
SpeculativeMethod = Literal["ngram", "eagle", "medusa", "mlp_speculator",
"draft_model"]
"draft_model", "deepseek_mtp"]
SpeculativeAcceptanceMethod = Literal["rejection_sampler",
"typical_acceptance_sampler"]
@@ -2519,6 +2519,15 @@ class SpeculativeConfig:
elif (self.draft_model_config.hf_config.model_type ==
"mlp_speculator"):
self.method = "mlp_speculator"
elif (self.draft_model_config.hf_config.model_type ==
"deepseek_mtp"):
self.method = "deepseek_mtp"
if self.num_speculative_tokens > 1:
logger.warning(
"All Deepseek MTP models only have " \
"one layer. Might need some code changes " \
"to support multiple layers."
)
else:
self.method = "draft_model"
@@ -2738,7 +2747,7 @@ class SpeculativeConfig:
return self.num_speculative_tokens
def use_eagle(self) -> bool:
return self.method in ("eagle", "eagle3")
return self.method in ("eagle", "eagle3", "deepseek_mtp")
def __repr__(self) -> str:
method = self.method