[Bugfix][Spec Decode] DSpark: inherit the target's attention backend when the speculative config names none (#52288)

2026-08-15 09:32:59 +08:00
parent 615d4cfade
commit acb0f1dcdb
1 changed files with 7 additions and 1 deletions
@@ -21,12 +21,18 @@ def load_dspark_model(target_model: nn.Module, vllm_config: VllmConfig) -> nn.Mo
    from vllm.model_executor.models.qwen3_dflash import dflash_has_any_non_causal
    from vllm.model_executor.models.utils import get_draft_quant_config

+    # None re-runs backend auto-selection for the draft, which can pick a
+    # different attention class than the target; fall back to the target's.
+    draft_attention_backend = (
+        speculative_config.attention_backend or vllm_config.attention_config.backend
+    )
+
    draft_vllm_config = replace(
        vllm_config,
        attention_config=replace(
            vllm_config.attention_config,
            use_non_causal=dflash_has_any_non_causal(draft_model_config.hf_config),
-            backend=speculative_config.attention_backend,
+            backend=draft_attention_backend,
        ),
        cache_config=(
            replace(