关联 Issue #162432 [standalone_compile] binary format write should be atomic #3149 Search-R1(lora) rollout outputs become garbled when applying LoRA with vLLM, but work fine without LoRA #31199 [Bug]: UnpicklingError during concurrent model compilation on multiple GPUs #35420 [Bugfix] Add monkeypatch to prevent race condition from writing
参与讨论