[Pooling] MRV2 pooling shutdown model ref (#57737)
Signed-off-by: Taneem Ibrahim <taneem.ibrahim@gmail.com>
Taneem Ibrahim committed
Sep 21, 2026 at 18:54 UTC
7d06dd1ffaab13bb6b9a34e86a8a62e4d715bd1e
1 file changed
+1
vllm/v1/worker/gpu/model_runner.py
+1
@@ -2245,6 +2245,7 @@ class GPUModelRunner(LoRAModelRunnerMixin):
2245
self.aux_output_connector.close()
2246
self.cudagraph_manager = None
2247
self.fast_prefill = None
2248
+ self.pooling_runner = None
2249
if hasattr(self, "kv_caches"):
2250
self.kv_caches.clear()
2251
if hasattr(self, "attn_groups"):