[2025-01-14 02:45:57,202][launcher][INFO] - ََAllocating process launcher [2025-01-14 02:45:57,202][process][INFO] - + Setting multiprocessing start method to spawn. [2025-01-14 02:45:57,220][process][INFO] - + Launched benchmark in isolated process 658. [PROC-0][2025-01-14 02:46:00,404][datasets][INFO] - PyTorch version 2.4.0 available. [PROC-0][2025-01-14 02:46:01,534][backend][INFO] - َAllocating pytorch backend [PROC-0][2025-01-14 02:46:01,534][backend][INFO] - + Setting random seed to 42 [PROC-0][2025-01-14 02:46:02,148][pytorch][INFO] - + Using AutoModel class AutoModelForCausalLM [PROC-0][2025-01-14 02:46:02,148][pytorch][INFO] - + Creating backend temporary directory [PROC-0][2025-01-14 02:46:02,148][pytorch][INFO] - + Loading model with random weights [PROC-0][2025-01-14 02:46:02,148][pytorch][INFO] - + Creating no weights model [PROC-0][2025-01-14 02:46:02,148][pytorch][INFO] - + Creating no weights model directory [PROC-0][2025-01-14 02:46:02,148][pytorch][INFO] - + Creating no weights model state dict [PROC-0][2025-01-14 02:46:02,151][pytorch][INFO] - + Saving no weights model safetensors [PROC-0][2025-01-14 02:46:02,151][pytorch][INFO] - + Saving no weights model pretrained config [PROC-0][2025-01-14 02:46:02,152][pytorch][INFO] - + Loading no weights AutoModel [PROC-0][2025-01-14 02:46:02,152][pytorch][INFO] - + Loading model directly on device: cuda [2025-01-14 02:46:03,353][experiment][ERROR] - Error during experiment