A high-throughput and memory-efficient inference and serving engine for LLMs
Mirrored from https://github.com/tduncan-lum/Nvidia-Optimization-Stack---VLLM.git
Synced Nov 11, 22:30 UTC
let draft model follow target model's config_format (#30152)
Bangsheng Tang committed
Dec 5, 2025 at 13:33 UTC
77e44728090cf11b6ece2ae25d7c732a065fbb45
1 file changed
+1
vllm/config/speculative.py
+1
@@ -337,6 +337,7 @@ class SpeculativeConfig:
337
enforce_eager=self.target_model_config.enforce_eager,
338
max_logprobs=self.target_model_config.max_logprobs,
339
hf_overrides=SpeculativeConfig.hf_config_override,
340
+ config_format=self.target_model_config.config_format,
341
)
342
343
# Automatically detect the method