SLURM_JOB_ID = 9168614 SLURM_JOB_NAME = nvr_elm_llm:dev/tw-8node-resume RUN_NAME = tw-8node-resume OUTPUT_DIR = runs/dev/tw-8node-resume NNODES = 8 NODES = batch-block1-3273 batch-block1-3339 batch-block1-3736 batch-block1-3544 batch-block1-1089 batch-block1-3578 batch-block1-1062 batch-block1-3035 NODE_RANK = 3 GPUS_PER_NODE = 8 MASTER_ADDR = batch-block1-3273 MASTER_PORT = 25001 GLOBAL_TRAIN_BATCH_SIZE = GRADIENT_ACCUMULATION_STEPS = PER_DEVICE_TRAIN_BATCH_SIZE = Resuming training from checkpoint: runs/dev/tw-8node-resume/checkpoint-64 SLURM_JOB_ID = 9168614 SLURM_JOB_NAME = nvr_elm_llm:dev/tw-8node-resume RUN_NAME = tw-8node-resume OUTPUT_DIR = runs/dev/tw-8node-resume NNODES = 8 SLURM_JOB_ID = 9168614 SLURM_JOB_NAME = nvr_elm_llm:dev/tw-8node-resume RUN_NAME = tw-8node-resume OUTPUT_DIR = runs/dev/tw-8node-resume NNODES = 8 SLURM_JOB_ID = 9168614 SLURM_JOB_NAME = nvr_elm_llm:dev/tw-8node-resume RUN_NAME = tw-8node-resume OUTPUT_DIR = runs/dev/tw-8node-resume NNODES = 8 SLURM_JOB_ID = 9168614 SLURM_JOB_NAME = nvr_elm_llm:dev/tw-8node-resume RUN_NAME = tw-8node-resume OUTPUT_DIR = runs/dev/tw-8node-resume NNODES = 8 SLURM_JOB_ID = 9168614 SLURM_JOB_NAME = nvr_elm_llm:dev/tw-8node-resume RUN_NAME = tw-8node-resume OUTPUT_DIR = runs/dev/tw-8node-resume NNODES = 8 SLURM_JOB_ID = 9168614 SLURM_JOB_NAME = nvr_elm_llm:dev/tw-8node-resume RUN_NAME = tw-8node-resume OUTPUT_DIR = runs/dev/tw-8node-resume NNODES = 8 SLURM_JOB_ID = 9168614 SLURM_JOB_NAME = nvr_elm_llm:dev/tw-8node-resume RUN_NAME = tw-8node-resume OUTPUT_DIR = runs/dev/tw-8node-resume NNODES = 8 NODES = batch-block1-3273 batch-block1-3339 batch-block1-3736 batch-block1-3544 batch-block1-1089 batch-block1-3578 batch-block1-1062 batch-block1-3035 NODE_RANK = 7 GPUS_PER_NODE = 8 NODES = batch-block1-3273 batch-block1-3339 batch-block1-3736 batch-block1-3544 batch-block1-1089 batch-block1-3578 batch-block1-1062 batch-block1-3035 NODE_RANK = 6 GPUS_PER_NODE = 8 NODES = batch-block1-3273 batch-block1-3339 batch-block1-3736 batch-block1-3544 batch-block1-1089 batch-block1-3578 batch-block1-1062 batch-block1-3035 NODE_RANK = 0 GPUS_PER_NODE = 8 NODES = batch-block1-3273 batch-block1-3339 batch-block1-3736 batch-block1-3544 batch-block1-1089 batch-block1-3578 batch-block1-1062 batch-block1-3035 NODE_RANK = 1 GPUS_PER_NODE = 8 NODES = batch-block1-3273 batch-block1-3339 batch-block1-3736 batch-block1-3544 batch-block1-1089 batch-block1-3578 batch-block1-1062 batch-block1-3035 NODE_RANK = 4 GPUS_PER_NODE = 8 NODES = batch-block1-3273 batch-block1-3339 batch-block1-3736 batch-block1-3544 batch-block1-1089 batch-block1-3578 batch-block1-1062 batch-block1-3035 NODE_RANK = 5 GPUS_PER_NODE = 8 MASTER_ADDR = batch-block1-3273 MASTER_PORT = 25001 GLOBAL_TRAIN_BATCH_SIZE = GRADIENT_ACCUMULATION_STEPS = PER_DEVICE_TRAIN_BATCH_SIZE = MASTER_ADDR = batch-block1-3273 MASTER_PORT = 25001 GLOBAL_TRAIN_BATCH_SIZE = GRADIENT_ACCUMULATION_STEPS = PER_DEVICE_TRAIN_BATCH_SIZE = NODES = batch-block1-3273 batch-block1-3339 batch-block1-3736 batch-block1-3544 batch-block1-1089 batch-block1-3578 batch-block1-1062 batch-block1-3035 NODE_RANK = 2 GPUS_PER_NODE = 8 MASTER_ADDR = batch-block1-3273 MASTER_PORT = 25001 GLOBAL_TRAIN_BATCH_SIZE = GRADIENT_ACCUMULATION_STEPS = PER_DEVICE_TRAIN_BATCH_SIZE = MASTER_ADDR = batch-block1-3273 MASTER_PORT = 25001 GLOBAL_TRAIN_BATCH_SIZE = GRADIENT_ACCUMULATION_STEPS = PER_DEVICE_TRAIN_BATCH_SIZE = MASTER_ADDR = batch-block1-3273 MASTER_ADDR = batch-block1-3273 MASTER_PORT = 25001 GLOBAL_TRAIN_BATCH_SIZE = GRADIENT_ACCUMULATION_STEPS = PER_DEVICE_TRAIN_BATCH_SIZE = MASTER_PORT = 25001 GLOBAL_TRAIN_BATCH_SIZE = GRADIENT_ACCUMULATION_STEPS = PER_DEVICE_TRAIN_BATCH_SIZE = Resuming training from checkpoint: runs/dev/tw-8node-resume/checkpoint-64 Resuming training from checkpoint: runs/dev/tw-8node-resume/checkpoint-64 Resuming training from checkpoint: runs/dev/tw-8node-resume/checkpoint-64 Resuming training from checkpoint: runs/dev/tw-8node-resume/checkpoint-64 Resuming training from checkpoint: runs/dev/tw-8node-resume/checkpoint-64 Resuming training from checkpoint: runs/dev/tw-8node-resume/checkpoint-64 MASTER_ADDR = batch-block1-3273 MASTER_PORT = 25001 GLOBAL_TRAIN_BATCH_SIZE = GRADIENT_ACCUMULATION_STEPS = PER_DEVICE_TRAIN_BATCH_SIZE = Resuming training from checkpoint: runs/dev/tw-8node-resume/checkpoint-64 Imported prefix tree collator v1 Imported prefix tree collator v1 Imported prefix tree collator v1 Imported prefix tree collator v1 Imported prefix tree collator v1 Imported prefix tree collator v1 Imported prefix tree collator v1 Imported prefix tree collator v1 Imported prefix tree collator v1Imported prefix tree collator v1 Imported prefix tree collator v1 Imported prefix tree collator v1 Imported prefix tree collator v1Imported prefix tree collator v1 Imported prefix tree collator v1 Imported prefix tree collator v1 Imported prefix tree collator v1Imported prefix tree collator v1 Imported prefix tree collator v1 Imported prefix tree collator v1 Imported prefix tree collator v1 Imported prefix tree collator v1 Imported prefix tree collator v1 Imported prefix tree collator v1 Imported prefix tree collator v1 Imported prefix tree collator v1 Imported prefix tree collator v1Imported prefix tree collator v1 Imported prefix tree collator v1 Imported prefix tree collator v1 Imported prefix tree collator v1 Imported prefix tree collator v1 Imported prefix tree collator v1Imported prefix tree collator v1 Imported prefix tree collator v1 Imported prefix tree collator v1 Imported prefix tree collator v1Imported prefix tree collator v1 Imported prefix tree collator v1 Imported prefix tree collator v1 Imported prefix tree collator v1 Imported prefix tree collator v1 Imported prefix tree collator v1 Imported prefix tree collator v1 Imported prefix tree collator v1 Imported prefix tree collator v1 Imported prefix tree collator v1 Imported prefix tree collator v1 Imported prefix tree collator v1 Imported prefix tree collator v1Imported prefix tree collator v1 Imported prefix tree collator v1Imported prefix tree collator v1 Imported prefix tree collator v1 Imported prefix tree collator v1 Imported prefix tree collator v1 Imported prefix tree collator v1 Imported prefix tree collator v1Imported prefix tree collator v1 Imported prefix tree collator v1Imported prefix tree collator v1 Imported prefix tree collator v1 Imported prefix tree collator v1 Imported prefix tree collator v1 [2026-04-13 22:52:44,641] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2026-04-13 22:52:44,641] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2026-04-13 22:52:44,641] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2026-04-13 22:52:44,641] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2026-04-13 22:52:44,641] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2026-04-13 22:52:44,641] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2026-04-13 22:52:44,641] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2026-04-13 22:52:44,641] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2026-04-13 22:52:44,823] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2026-04-13 22:52:44,823] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2026-04-13 22:52:44,823] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2026-04-13 22:52:44,823] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2026-04-13 22:52:44,823] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2026-04-13 22:52:44,823] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2026-04-13 22:52:44,823] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2026-04-13 22:52:44,824] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2026-04-13 22:52:44,830] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2026-04-13 22:52:44,830] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2026-04-13 22:52:44,830] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2026-04-13 22:52:44,830] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2026-04-13 22:52:44,830] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2026-04-13 22:52:44,830] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2026-04-13 22:52:44,830] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2026-04-13 22:52:44,831] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2026-04-13 22:52:45,044] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2026-04-13 22:52:45,044] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2026-04-13 22:52:45,044] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2026-04-13 22:52:45,044] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2026-04-13 22:52:45,044] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2026-04-13 22:52:45,044] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2026-04-13 22:52:45,044] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2026-04-13 22:52:45,044] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2026-04-13 22:52:45,505] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2026-04-13 22:52:45,505] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2026-04-13 22:52:45,505] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2026-04-13 22:52:45,505] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2026-04-13 22:52:45,505] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2026-04-13 22:52:45,505] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2026-04-13 22:52:45,505] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2026-04-13 22:52:45,505] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2026-04-13 22:52:45,710] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2026-04-13 22:52:45,710] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2026-04-13 22:52:45,710] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2026-04-13 22:52:45,710] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2026-04-13 22:52:45,710] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2026-04-13 22:52:45,710] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2026-04-13 22:52:45,710] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2026-04-13 22:52:45,711] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2026-04-13 22:52:46,098] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2026-04-13 22:52:46,098] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2026-04-13 22:52:46,098] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2026-04-13 22:52:46,098] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2026-04-13 22:52:46,098] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2026-04-13 22:52:46,098] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2026-04-13 22:52:46,098] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2026-04-13 22:52:46,099] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2026-04-13 22:52:46,114] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2026-04-13 22:52:46,114] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2026-04-13 22:52:46,114] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2026-04-13 22:52:46,114] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2026-04-13 22:52:46,114] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2026-04-13 22:52:46,114] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2026-04-13 22:52:46,114] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2026-04-13 22:52:46,115] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect) [2026-04-13 22:52:57,932] [INFO] [comm.py:675:init_distributed] cdb=None [2026-04-13 22:52:57,932] [INFO] [comm.py:675:init_distributed] cdb=None [2026-04-13 22:52:57,932] [INFO] [comm.py:675:init_distributed] cdb=None [2026-04-13 22:52:57,932] [INFO] [comm.py:675:init_distributed] cdb=None [2026-04-13 22:52:57,932] [INFO] [comm.py:675:init_distributed] cdb=None [2026-04-13 22:52:57,932] [INFO] [comm.py:675:init_distributed] cdb=None [2026-04-13 22:52:57,932] [INFO] [comm.py:675:init_distributed] cdb=None [2026-04-13 22:52:57,932] [INFO] [comm.py:675:init_distributed] cdb=None [2026-04-13 22:52:58,617] [INFO] [comm.py:675:init_distributed] cdb=None [2026-04-13 22:52:58,617] [INFO] [comm.py:675:init_distributed] cdb=None [2026-04-13 22:52:58,617] [INFO] [comm.py:675:init_distributed] cdb=None [2026-04-13 22:52:58,617] [INFO] [comm.py:675:init_distributed] cdb=None [2026-04-13 22:52:58,617] [INFO] [comm.py:675:init_distributed] cdb=None [2026-04-13 22:52:58,617] [INFO] [comm.py:675:init_distributed] cdb=None [2026-04-13 22:52:58,618] [INFO] [comm.py:675:init_distributed] cdb=None [2026-04-13 22:52:58,618] [INFO] [comm.py:675:init_distributed] cdb=None [2026-04-13 22:52:58,976] [INFO] [comm.py:675:init_distributed] cdb=None [2026-04-13 22:52:58,976] [INFO] [comm.py:675:init_distributed] cdb=None [2026-04-13 22:52:58,976] [INFO] [comm.py:675:init_distributed] cdb=None [2026-04-13 22:52:58,976] [INFO] [comm.py:706:init_distributed] Initializing TorchBackend in DeepSpeed with backend nccl [2026-04-13 22:52:58,976] [INFO] [comm.py:675:init_distributed] cdb=None [2026-04-13 22:52:58,977] [INFO] [comm.py:675:init_distributed] cdb=None [2026-04-13 22:52:58,977] [INFO] [comm.py:675:init_distributed] cdb=None [2026-04-13 22:52:58,976] [INFO] [comm.py:675:init_distributed] cdb=None [2026-04-13 22:52:58,977] [INFO] [comm.py:675:init_distributed] cdb=None [2026-04-13 22:52:58,977] [INFO] [comm.py:675:init_distributed] cdb=None [2026-04-13 22:52:58,977] [INFO] [comm.py:675:init_distributed] cdb=None [2026-04-13 22:52:58,977] [INFO] [comm.py:675:init_distributed] cdb=None [2026-04-13 22:52:58,977] [INFO] [comm.py:675:init_distributed] cdb=None [2026-04-13 22:52:58,977] [INFO] [comm.py:675:init_distributed] cdb=None [2026-04-13 22:52:58,978] [INFO] [comm.py:675:init_distributed] cdb=None [2026-04-13 22:52:58,978] [INFO] [comm.py:675:init_distributed] cdb=None [2026-04-13 22:52:58,978] [INFO] [comm.py:675:init_distributed] cdb=None [2026-04-13 22:52:59,398] [INFO] [comm.py:675:init_distributed] cdb=None [2026-04-13 22:52:59,398] [INFO] [comm.py:675:init_distributed] cdb=None [2026-04-13 22:52:59,398] [INFO] [comm.py:675:init_distributed] cdb=None [2026-04-13 22:52:59,399] [INFO] [comm.py:675:init_distributed] cdb=None [2026-04-13 22:52:59,399] [INFO] [comm.py:675:init_distributed] cdb=None [2026-04-13 22:52:59,399] [INFO] [comm.py:675:init_distributed] cdb=None [2026-04-13 22:52:59,399] [INFO] [comm.py:675:init_distributed] cdb=None [2026-04-13 22:52:59,399] [INFO] [comm.py:675:init_distributed] cdb=None [2026-04-13 22:52:59,650] [INFO] [comm.py:675:init_distributed] cdb=None [2026-04-13 22:52:59,650] [INFO] [comm.py:675:init_distributed] cdb=None [2026-04-13 22:52:59,650] [INFO] [comm.py:675:init_distributed] cdb=None [2026-04-13 22:52:59,650] [INFO] [comm.py:675:init_distributed] cdb=None [2026-04-13 22:52:59,650] [INFO] [comm.py:675:init_distributed] cdb=None [2026-04-13 22:52:59,650] [INFO] [comm.py:675:init_distributed] cdb=None [2026-04-13 22:52:59,650] [INFO] [comm.py:675:init_distributed] cdb=None [2026-04-13 22:52:59,650] [INFO] [comm.py:675:init_distributed] cdb=None [2026-04-13 22:52:59,893] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64 [2026-04-13 22:52:59,894] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64 [2026-04-13 22:52:59,924] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64 [2026-04-13 22:52:59,924] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64 [2026-04-13 22:52:59,924] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64 [2026-04-13 22:52:59,988] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64 [2026-04-13 22:52:59,988] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64 [2026-04-13 22:52:59,988] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64 [2026-04-13 22:52:59,988] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64 [2026-04-13 22:52:59,988] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64 [2026-04-13 22:52:59,988] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64 [2026-04-13 22:52:59,988] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64 [2026-04-13 22:52:59,988] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64 [2026-04-13 22:52:59,988] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64 [2026-04-13 22:52:59,988] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64 [2026-04-13 22:52:59,988] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64 [2026-04-13 22:52:59,989] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64 [2026-04-13 22:53:00,251] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64 [2026-04-13 22:53:00,305] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64 [2026-04-13 22:53:00,315] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64 [2026-04-13 22:53:00,318] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64 [2026-04-13 22:53:00,318] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64 [2026-04-13 22:53:00,320] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64 [2026-04-13 22:53:00,320] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64 [2026-04-13 22:53:00,320] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64 [2026-04-13 22:53:00,321] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64 [2026-04-13 22:53:00,332] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64 [2026-04-13 22:53:00,333] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64 [2026-04-13 22:53:00,333] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64 [2026-04-13 22:53:00,337] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64 [2026-04-13 22:53:00,338] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64 [2026-04-13 22:53:00,338] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64 [2026-04-13 22:53:00,342] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64 [2026-04-13 22:53:00,383] [INFO] [comm.py:675:init_distributed] cdb=None [2026-04-13 22:53:00,383] [INFO] [comm.py:675:init_distributed] cdb=None [2026-04-13 22:53:00,383] [INFO] [comm.py:675:init_distributed] cdb=None [2026-04-13 22:53:00,383] [INFO] [comm.py:675:init_distributed] cdb=None [2026-04-13 22:53:00,383] [INFO] [comm.py:675:init_distributed] cdb=None [2026-04-13 22:53:00,383] [INFO] [comm.py:675:init_distributed] cdb=None [2026-04-13 22:53:00,383] [INFO] [comm.py:675:init_distributed] cdb=None [2026-04-13 22:53:00,384] [INFO] [comm.py:675:init_distributed] cdb=None [2026-04-13 22:53:00,471] [INFO] [comm.py:675:init_distributed] cdb=None [2026-04-13 22:53:00,471] [INFO] [comm.py:675:init_distributed] cdb=None [2026-04-13 22:53:00,471] [INFO] [comm.py:675:init_distributed] cdb=None [2026-04-13 22:53:00,471] [INFO] [comm.py:675:init_distributed] cdb=None [2026-04-13 22:53:00,472] [INFO] [comm.py:675:init_distributed] cdb=None [2026-04-13 22:53:00,472] [INFO] [comm.py:675:init_distributed] cdb=None [2026-04-13 22:53:00,472] [INFO] [comm.py:675:init_distributed] cdb=None [2026-04-13 22:53:00,472] [INFO] [comm.py:675:init_distributed] cdb=None [2026-04-13 22:53:02,594] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64 [2026-04-13 22:53:02,604] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64 [2026-04-13 22:53:02,604] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64 [2026-04-13 22:53:02,627] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64 [2026-04-13 22:53:02,628] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64 [2026-04-13 22:53:02,634] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64 [2026-04-13 22:53:02,634] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64 [2026-04-13 22:53:02,645] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64 [2026-04-13 22:53:02,672] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64 [2026-04-13 22:53:02,672] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64 [2026-04-13 22:53:02,672] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64 [2026-04-13 22:53:02,672] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64 [2026-04-13 22:53:02,675] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64 [2026-04-13 22:53:02,680] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64 [2026-04-13 22:53:02,692] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64 [2026-04-13 22:53:02,693] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64 [2026-04-13 22:53:05,737] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64 [2026-04-13 22:53:05,750] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64 [2026-04-13 22:53:05,779] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64 [2026-04-13 22:53:05,813] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64 [2026-04-13 22:53:05,854] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64 [2026-04-13 22:53:05,858] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64 [2026-04-13 22:53:05,885] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64 [2026-04-13 22:53:05,899] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64 [2026-04-13 22:53:05,899] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64 [2026-04-13 22:53:05,917] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64 [2026-04-13 22:53:05,932] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64 [2026-04-13 22:53:05,949] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64 [2026-04-13 22:53:05,959] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64 [2026-04-13 22:53:05,959] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64 [2026-04-13 22:53:05,961] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64 [2026-04-13 22:53:10,377] [INFO] [partition_parameters.py:348:__exit__] finished initializing model - num_params = 399, num_elems = 8.19B Using Qwen templates for runs/dev/tw-8node-resume/checkpoint-64Using Qwen templates for runs/dev/tw-8node-resume/checkpoint-64 Using Qwen templates for runs/dev/tw-8node-resume/checkpoint-64 Using Qwen templates for runs/dev/tw-8node-resume/checkpoint-64 Using Qwen templates for runs/dev/tw-8node-resume/checkpoint-64Using Qwen templates for runs/dev/tw-8node-resume/checkpoint-64 Using Qwen templates for runs/dev/tw-8node-resume/checkpoint-64Using Qwen templates for runs/dev/tw-8node-resume/checkpoint-64 Using Qwen templates for runs/dev/tw-8node-resume/checkpoint-64 Using Qwen templates for runs/dev/tw-8node-resume/checkpoint-64Using Qwen templates for runs/dev/tw-8node-resume/checkpoint-64 Using Qwen templates for runs/dev/tw-8node-resume/checkpoint-64 Using Qwen templates for runs/dev/tw-8node-resume/checkpoint-64 Using Qwen templates for runs/dev/tw-8node-resume/checkpoint-64 Using Qwen templates for runs/dev/tw-8node-resume/checkpoint-64Using Qwen templates for runs/dev/tw-8node-resume/checkpoint-64 Using Qwen templates for runs/dev/tw-8node-resume/checkpoint-64 Using Qwen templates for runs/dev/tw-8node-resume/checkpoint-64 Using Qwen templates for runs/dev/tw-8node-resume/checkpoint-64 Using Qwen templates for runs/dev/tw-8node-resume/checkpoint-64 Using Qwen templates for runs/dev/tw-8node-resume/checkpoint-64 Using Qwen templates for runs/dev/tw-8node-resume/checkpoint-64 Using Qwen templates for runs/dev/tw-8node-resume/checkpoint-64 Using Qwen templates for runs/dev/tw-8node-resume/checkpoint-64 Using Qwen templates for runs/dev/tw-8node-resume/checkpoint-64 Using Qwen templates for runs/dev/tw-8node-resume/checkpoint-64 Using Qwen templates for runs/dev/tw-8node-resume/checkpoint-64 Using Qwen templates for runs/dev/tw-8node-resume/checkpoint-64 Using Qwen templates for runs/dev/tw-8node-resume/checkpoint-64 Using Qwen templates for runs/dev/tw-8node-resume/checkpoint-64 Using Qwen templates for runs/dev/tw-8node-resume/checkpoint-64 Using Qwen templates for runs/dev/tw-8node-resume/checkpoint-64Using Qwen templates for runs/dev/tw-8node-resume/checkpoint-64 Using Qwen templates for runs/dev/tw-8node-resume/checkpoint-64 Using Qwen templates for runs/dev/tw-8node-resume/checkpoint-64 Using Qwen templates for runs/dev/tw-8node-resume/checkpoint-64 Using Qwen templates for runs/dev/tw-8node-resume/checkpoint-64 Using Qwen templates for runs/dev/tw-8node-resume/checkpoint-64 Using Qwen templates for runs/dev/tw-8node-resume/checkpoint-64 Using Qwen templates for runs/dev/tw-8node-resume/checkpoint-64 Using Qwen templates for runs/dev/tw-8node-resume/checkpoint-64 Using Qwen templates for runs/dev/tw-8node-resume/checkpoint-64 Using Qwen templates for runs/dev/tw-8node-resume/checkpoint-64 Using Qwen templates for runs/dev/tw-8node-resume/checkpoint-64 Using Qwen templates for runs/dev/tw-8node-resume/checkpoint-64 Using Qwen templates for runs/dev/tw-8node-resume/checkpoint-64 Using Qwen templates for runs/dev/tw-8node-resume/checkpoint-64 Using Qwen templates for runs/dev/tw-8node-resume/checkpoint-64 Using Qwen templates for runs/dev/tw-8node-resume/checkpoint-64 Using Qwen templates for runs/dev/tw-8node-resume/checkpoint-64 Using Qwen templates for runs/dev/tw-8node-resume/checkpoint-64 Using Qwen templates for runs/dev/tw-8node-resume/checkpoint-64 Using Qwen templates for runs/dev/tw-8node-resume/checkpoint-64 Using Qwen templates for runs/dev/tw-8node-resume/checkpoint-64 Using Qwen templates for runs/dev/tw-8node-resume/checkpoint-64 Using Qwen templates for runs/dev/tw-8node-resume/checkpoint-64 Using Qwen templates for runs/dev/tw-8node-resume/checkpoint-64 Using Qwen templates for runs/dev/tw-8node-resume/checkpoint-64 Using Qwen templates for runs/dev/tw-8node-resume/checkpoint-64 Using Qwen templates for runs/dev/tw-8node-resume/checkpoint-64 Using Qwen templates for runs/dev/tw-8node-resume/checkpoint-64 Using Qwen templates for runs/dev/tw-8node-resume/checkpoint-64 Using Qwen templates for runs/dev/tw-8node-resume/checkpoint-64 Using Qwen templates for runs/dev/tw-8node-resume/checkpoint-64 Using Prefix Tree collator args.report_to: ['wandb'] args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-52-59_batch-block1-3578Using Prefix Tree collator Using Prefix Tree collator Using Prefix Tree collator Using Prefix Tree collator Using Prefix Tree collator Using Prefix Tree collator args.report_to: ['wandb']Using Prefix Tree collator args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-52-59_batch-block1-3578 args.report_to: ['wandb'] args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-52-59_batch-block1-3578 args.report_to: ['wandb'] args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-52-59_batch-block1-3578 args.report_to: ['wandb'] args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-52-59_batch-block1-3578 args.report_to: ['wandb'] args.report_to: ['wandb']args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-52-59_batch-block1-3578 args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-52-59_batch-block1-3578 args.report_to: ['wandb'] args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-52-59_batch-block1-3578 Using Prefix Tree collatorUsing Prefix Tree collatorUsing Prefix Tree collator Using Prefix Tree collator Using Prefix Tree collator Using Prefix Tree collator Using Prefix Tree collator Using Prefix Tree collator args.report_to: ['wandb']args.report_to: ['wandb'] args.report_to: ['wandb']args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-52-58_batch-block1-1062args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-52-58_batch-block1-1062 args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-52-58_batch-block1-1062 args.report_to: ['wandb'] args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-52-58_batch-block1-1062 args.report_to: ['wandb'] args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-52-58_batch-block1-1062 args.report_to: ['wandb'] args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-52-58_batch-block1-1062 args.report_to: ['wandb'] args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-52-58_batch-block1-1062 args.report_to: ['wandb'] args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-52-58_batch-block1-1062 Using Prefix Tree collator Using Prefix Tree collator Using Prefix Tree collatorUsing Prefix Tree collator Using Prefix Tree collator Using Prefix Tree collator Using Prefix Tree collatorUsing Prefix Tree collator args.report_to: ['wandb'] args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-53-00_batch-block1-3736 args.report_to: ['wandb'] args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-53-00_batch-block1-3736 args.report_to: ['wandb']args.report_to: ['wandb'] args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-53-00_batch-block1-3736args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-53-00_batch-block1-3736 args.report_to: ['wandb']args.report_to: ['wandb'] args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-53-00_batch-block1-3736args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-53-00_batch-block1-3736 args.report_to: ['wandb'] args.report_to: ['wandb']args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-53-00_batch-block1-3736 args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-53-00_batch-block1-3736 Using Prefix Tree collator Using Prefix Tree collatorUsing Prefix Tree collator Using Prefix Tree collatorUsing Prefix Tree collator Using Prefix Tree collator Using Prefix Tree collator Using Prefix Tree collator args.report_to: ['wandb'] args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-53-00_batch-block1-1089 args.report_to: ['wandb'] args.report_to: ['wandb']args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-53-00_batch-block1-1089 args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-53-00_batch-block1-1089 args.report_to: ['wandb']args.report_to: ['wandb'] args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-53-00_batch-block1-1089args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-53-00_batch-block1-1089 args.report_to: ['wandb'] args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-53-00_batch-block1-1089 args.report_to: ['wandb'] args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-53-00_batch-block1-1089 args.report_to: ['wandb'] args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-53-00_batch-block1-1089 Using Prefix Tree collator Using Prefix Tree collatorUsing Prefix Tree collator args.report_to: ['wandb'] args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-52-58_batch-block1-3273 Using Prefix Tree collator args.report_to: ['wandb'] args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-52-58_batch-block1-3273 args.report_to: ['wandb'] args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-52-58_batch-block1-3273 Using Prefix Tree collator args.report_to: ['wandb'] args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-52-58_batch-block1-3273 Using Prefix Tree collatorargs.report_to: ['wandb'] args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-52-58_batch-block1-3273 Using Prefix Tree collator Using Prefix Tree collator args.report_to: ['wandb'] args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-52-58_batch-block1-3273 args.report_to: ['wandb'] args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-52-58_batch-block1-3273 args.report_to: ['wandb'] args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-52-58_batch-block1-3273 Using Prefix Tree collatorUsing Prefix Tree collator Using Prefix Tree collator Using Prefix Tree collator Using Prefix Tree collator Using Prefix Tree collator args.report_to: ['wandb']args.report_to: ['wandb']args.report_to: ['wandb'] args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-52-59_batch-block1-3339args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-52-59_batch-block1-3339 args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-52-59_batch-block1-3339 args.report_to: ['wandb'] args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-52-59_batch-block1-3339 args.report_to: ['wandb'] args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-52-59_batch-block1-3339 args.report_to: ['wandb'] args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-52-59_batch-block1-3339 Using Prefix Tree collator Using Prefix Tree collator args.report_to: ['wandb'] args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-52-59_batch-block1-3339args.report_to: ['wandb'] args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-52-59_batch-block1-3339 Using Prefix Tree collator Using Prefix Tree collator Using Prefix Tree collatorUsing Prefix Tree collator Using Prefix Tree collator args.report_to: ['wandb'] args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-52-58_batch-block1-3035 args.report_to: ['wandb'] args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-52-58_batch-block1-3035 args.report_to: ['wandb'] args.report_to: ['wandb']args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-52-58_batch-block1-3035 args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-52-58_batch-block1-3035 Using Prefix Tree collator args.report_to: ['wandb']Using Prefix Tree collator args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-52-58_batch-block1-3035 Using Prefix Tree collator args.report_to: ['wandb'] args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-52-58_batch-block1-3035 args.report_to: ['wandb'] args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-52-58_batch-block1-3035 args.report_to: ['wandb'] args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-52-58_batch-block1-3035 Using Prefix Tree collator Using Prefix Tree collatorUsing Prefix Tree collatorUsing Prefix Tree collator Using Prefix Tree collator Using Prefix Tree collator Using Prefix Tree collator Using Prefix Tree collator args.report_to: ['wandb'] args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-52-57_batch-block1-3544 args.report_to: ['wandb']args.report_to: ['wandb'] args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-52-57_batch-block1-3544args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-52-57_batch-block1-3544 args.report_to: ['wandb'] args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-52-57_batch-block1-3544 args.report_to: ['wandb'] args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-52-57_batch-block1-3544 args.report_to: ['wandb'] args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-52-57_batch-block1-3544 args.report_to: ['wandb']args.report_to: ['wandb'] args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-52-57_batch-block1-3544args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-52-57_batch-block1-3544 Parameter Offload: Total persistent parameters: 308224 in 145 params {'train_runtime': 4.1111, 'train_samples_per_second': 1875.899, 'train_steps_per_second': 15.568, 'train_loss': 0.0, 'epoch': 8.0} wandb: wandb: 🚀 View run runs/dev/tw-8node-resume at: https://wandb.ai/ligeng-zhu/ThreadWeaver/runs/tw-8node-resume