[Bugfix][HiSparse] Fix MTP acceptance collapse under FULL graphs with a saturated GPU pool (#59309)

Signed-off-by: Lucas Wilkinson <lwilkinson@neuralmagic.com>
Signed-off-by: Matthew Bonanni <mbonanni@redhat.com>
Co-authored-by: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Co-authored-by: Matthew Bonanni <mbonanni@redhat.com>
This commit is contained in:
Lucas Wilkinson
2026-10-01 23:04:21 -07:00
committed by GitHub
co-authored by Claude Opus 5.5 Matthew Bonanni
parent 0cbac6cd13
commit 4056c8ac1f
11 changed files with 426 additions and 122 deletions
+6 -6
View File
@@ -224,13 +224,13 @@ class DraftModelSpeculator(BaseSpeculator):
self.model = self.load_draft_model(target_model, target_attn_layer_names)
self._validate_local_argmax_reduction()
all_attn_layers = set[str](
get_layers_from_vllm_config(
self.vllm_config,
AttentionLayerBase, # type: ignore[type-abstract]
).keys()
all_attn_layers = get_layers_from_vllm_config(
self.vllm_config,
AttentionLayerBase, # type: ignore[type-abstract]
)
self.draft_attn_layer_names = all_attn_layers - target_attn_layer_names
self.draft_attn_layer_names = set(all_attn_layers) - target_attn_layer_names
for layer_name in self.draft_attn_layer_names:
all_attn_layers[layer_name].is_draft_layer = True
target_supports_mm = self.vllm_config.model_config.supports_multimodal_inputs
draft_supports_mm = supports_multimodal_embeddings(self.model)