W0624 23:17:56.656000 3086581 site-packages/torch/distributed/run.py:766] W0624 23:17:56.656000 3086581 site-packages/torch/distributed/run.py:766] ***************************************** W0624 23:17:56.656000 3086581 site-packages/torch/distributed/run.py:766] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. W0624 23:17:56.656000 3086581 site-packages/torch/distributed/run.py:766] ***************************************** [INFO] Running in WANDB offline mode [INFO] Running in WANDB offline mode [INFO] Running in WANDB offline mode [INFO] Running in WANDB offline mode Trainer._get_train_sampler replaced with custom implementation. warmup_ratio is deprecated and will be removed in v5.2. Use `warmup_steps` instead. [2026-06-24 23:18:03,725] [INFO] [real_accelerator.py:222:get_accelerator] Setting ds_accelerator to cuda (auto detect) df: /home/dungpq6/.triton/autotune: No such file or directory Trainer._get_train_sampler replaced with custom implementation. warmup_ratio is deprecated and will be removed in v5.2. Use `warmup_steps` instead. [2026-06-24 23:18:04,045] [INFO] [real_accelerator.py:222:get_accelerator] Setting ds_accelerator to cuda (auto detect) Trainer._get_train_sampler replaced with custom implementation. [2026-06-24 23:18:04,086] [INFO] [comm.py:658:init_distributed] cdb=None warmup_ratio is deprecated and will be removed in v5.2. Use `warmup_steps` instead. Trainer._get_train_sampler replaced with custom implementation. warmup_ratio is deprecated and will be removed in v5.2. Use `warmup_steps` instead. [2026-06-24 23:18:04,136] [INFO] [real_accelerator.py:222:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2026-06-24 23:18:04,159] [INFO] [real_accelerator.py:222:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2026-06-24 23:18:04,349] [INFO] [comm.py:658:init_distributed] cdb=None [2026-06-24 23:18:04,500] [INFO] [comm.py:658:init_distributed] cdb=None [2026-06-24 23:18:04,507] [INFO] [comm.py:658:init_distributed] cdb=None [2026-06-24 23:18:04,507] [INFO] [comm.py:689:init_distributed] Initializing TorchBackend in DeepSpeed with backend nccl Loading weights: 0%| | 0/723 [00:00