[WIP] Before modify nanovllm CPU-GPU kvcache.

2025-12-31 22:41:07 +08:00
parent 31e90a7268
commit ccd1b3d4ab
2 changed files with 47 additions and 193 deletions
--- a/nanovllm/kvcache/offload_engine.py
+++ b/nanovllm/kvcache/offload_engine.py
@@ -1093,4 +1093,7 @@ class OffloadEngine:
            try:
                hook(slot_idx, layer_id, cpu_block_id, k, v)
            except Exception as e:
+                # Allow pdb quit to propagate
+                if e.__class__.__name__ == 'BdbQuit':
+                    raise
                logger.warning(f"Debug hook error: {e}")