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 15, 07:15 UTC
[Bugfix][Model] Attempt to fix eagle in V0. (#18978)
Signed-off-by: Gregory Shtrasberg <Gregory.Shtrasberg@amd.com>
Gregory Shtrasberg committed
Jun 2, 2025 at 11:15 UTC
ca2f6b9c301df6dbe2e5c83c705051f478140695
1 file changed
+2
-1
vllm/config.py
+2
-1
@@ -2595,7 +2595,8 @@ class SpeculativeConfig:
2595
else:
2596
eagle_config = EAGLEConfig(
2597
self.draft_model_config.hf_config,
2598
- method=self.method)
2598
+ method=self.method,
2599
+ model_type="eagle")
2600
self.draft_model_config.hf_config = eagle_config
2601
2602
if (self.num_speculative_tokens is not None