iHore/vLLM
8
0
Fork 0
Code Issues Pull Requests Actions 1 Packages Projects Releases Wiki Activity

[BugFix][SpecDecode] Fix dspark parallel_drafting_token_id init bug (#51602)

Browse Source 
Signed-off-by: wangxiyuan <wangxiyuan1007@gmail.com>
Co-authored-by: Benjamin Chislett <bchislett@nvidia.com>
This commit is contained in:
wangxiyuan
2026-08-10 13:33:10 -04:00
committed by GitHub
co-authored by Benjamin Chislett
parent 3a749ce816
commit 355a338b8f
1 changed files with 7 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files
vllm/v1/spec_decode/llm_base_proposer.py
+7 -2
View File
@@ -357,6 +357,10 @@ class SpecDecodeBaseProposer:
dflash_config = getattr(model_hf_config, "dflash_config", None)
if dflash_config and "mask_token_id" in dflash_config:
self.parallel_drafting_token_id = dflash_config["mask_token_id"]
elif getattr(model_hf_config, "mask_token_id", None) is not None:
self.parallel_drafting_token_id = model_hf_config.mask_token_id
elif hasattr(model_hf_config, "dspark_noise_token_id"):
self.parallel_drafting_token_id = model_hf_config.dspark_noise_token_id
elif hasattr(model_hf_config, "pard_token"):
self.parallel_drafting_token_id = model_hf_config.pard_token
elif hasattr(model_hf_config, "ptd_token_id"):
@@ -364,8 +368,9 @@ class SpecDecodeBaseProposer:
else:
raise ValueError(
"For parallel drafting, the draft model config must have "
"`pard_token`, `ptd_token_id`, or "
"`dflash_config.mask_token_id` specified in its config.json."
"`dflash_config.mask_token_id`, `mask_token_id`, "
"`dspark_noise_token_id`, `pard_token`, or `ptd_token_id` "
"specified in its config.json."
)
if self.pass_hidden_states_to_model: