[Fix] missing max_context_len on HybridAttnBackend (#32690)

2026-07-31 19:43:09 +08:00
parent 754b692afc
commit e3d4f48e55
4 changed files with 4 additions and 0 deletions
@@ -42,6 +42,7 @@ class HybridAttnBackend(AttentionBackend):
        self.needs_cpu_seq_lens = (
            prefill_backend.needs_cpu_seq_lens or decode_backend.needs_cpu_seq_lens
        )
+        self.max_context_len = model_runner.model_config.context_len

    def _select_backend(self, forward_mode: ForwardMode) -> AttentionBackend:
        """