Files
q3-8b-train_final_v2_nb2_mt…/slurm/9168624.0.out

631 lines
38 KiB
Plaintext
Raw Normal View History

SLURM_JOB_ID = 9168624
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 = 9168624
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-block1-3736 batch-block1-3305 batch-block1-3544 batch-block1-1089 batch-block1-3163 batch-block1-3038 batch-block1-1062 batch-block1-3035
NODE_RANK = 4
GPUS_PER_NODE = 8
NODES = batch-block1-3736 batch-block1-3305 batch-block1-3544 batch-block1-1089 batch-block1-3163 batch-block1-3038 batch-block1-1062 batch-block1-3035
NODE_RANK = 5
GPUS_PER_NODE = 8
MASTER_ADDR = batch-block1-3736
MASTER_PORT = 25001
GLOBAL_TRAIN_BATCH_SIZE =
GRADIENT_ACCUMULATION_STEPS =
PER_DEVICE_TRAIN_BATCH_SIZE =
MASTER_ADDR = batch-block1-3736
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
SLURM_JOB_ID = 9168624
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 = 9168624
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 = 9168624
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 = 9168624
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-block1-3736 batch-block1-3305 batch-block1-3544 batch-block1-1089 batch-block1-3163 batch-block1-3038 batch-block1-1062 batch-block1-3035
NODE_RANK = 7
GPUS_PER_NODE = 8
NODES = batch-block1-3736 batch-block1-3305 batch-block1-3544 batch-block1-1089 batch-block1-3163 batch-block1-3038 batch-block1-1062 batch-block1-3035
NODE_RANK = 3
GPUS_PER_NODE = 8
SLURM_JOB_ID = 9168624
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-block1-3736 batch-block1-3305 batch-block1-3544 batch-block1-1089 batch-block1-3163 batch-block1-3038 batch-block1-1062 batch-block1-3035
NODE_RANK = 1
GPUS_PER_NODE = 8
SLURM_JOB_ID = 9168624
SLURM_JOB_NAME = nvr_elm_llm:dev/tw-8node-resume
RUN_NAME = tw-8node-resume
OUTPUT_DIR = runs/dev/tw-8node-resume
NNODES = 8
MASTER_ADDR = batch-block1-3736
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
NODES = batch-block1-3736 batch-block1-3305 batch-block1-3544 batch-block1-1089 batch-block1-3163 batch-block1-3038 batch-block1-1062 batch-block1-3035
NODE_RANK = 2
GPUS_PER_NODE = 8
MASTER_ADDR = batch-block1-3736
MASTER_PORT = 25001
GLOBAL_TRAIN_BATCH_SIZE =
GRADIENT_ACCUMULATION_STEPS =
PER_DEVICE_TRAIN_BATCH_SIZE =
NODES = batch-block1-3736 batch-block1-3305 batch-block1-3544 batch-block1-1089 batch-block1-3163 batch-block1-3038 batch-block1-1062 batch-block1-3035
NODE_RANK = 0
GPUS_PER_NODE = 8
MASTER_ADDR = batch-block1-3736
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
NODES = batch-block1-3736 batch-block1-3305 batch-block1-3544 batch-block1-1089 batch-block1-3163 batch-block1-3038 batch-block1-1062 batch-block1-3035
NODE_RANK = 6
GPUS_PER_NODE = 8
Resuming training from checkpoint: runs/dev/tw-8node-resume/checkpoint-64
MASTER_ADDR = batch-block1-3736
MASTER_PORT = 25001
GLOBAL_TRAIN_BATCH_SIZE =
GRADIENT_ACCUMULATION_STEPS =
PER_DEVICE_TRAIN_BATCH_SIZE =
MASTER_ADDR = batch-block1-3736
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
MASTER_ADDR = batch-block1-3736
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
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 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 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 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 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 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 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
[2026-04-13 22:57:05,613] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect)
[2026-04-13 22:57:05,613] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect)
[2026-04-13 22:57:05,675] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect)
[2026-04-13 22:57:05,705] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect)
[2026-04-13 22:57:05,869] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect)
[2026-04-13 22:57:05,871] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect)
[2026-04-13 22:57:05,874] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect)
[2026-04-13 22:57:05,876] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect)
[2026-04-13 22:57:05,882] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect)
[2026-04-13 22:57:05,885] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect)
[2026-04-13 22:57:05,902] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect)
[2026-04-13 22:57:05,962] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect)
[2026-04-13 22:57:05,967] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect)
[2026-04-13 22:57:05,997] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect)
[2026-04-13 22:57:06,005] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect)
[2026-04-13 22:57:06,008] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect)
[2026-04-13 22:57:06,011] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect)
[2026-04-13 22:57:06,016] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect)
[2026-04-13 22:57:06,041] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect)
[2026-04-13 22:57:06,048] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect)
[2026-04-13 22:57:06,055] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect)
[2026-04-13 22:57:06,076] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect)
[2026-04-13 22:57:06,078] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect)
[2026-04-13 22:57:06,108] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect)
[2026-04-13 22:57:06,119] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect)
[2026-04-13 22:57:06,122] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect)
[2026-04-13 22:57:06,280] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect)
[2026-04-13 22:57:06,346] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect)
[2026-04-13 22:57:06,401] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect)
[2026-04-13 22:57:06,418] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect)
[2026-04-13 22:57:06,442] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect)
[2026-04-13 22:57:06,456] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect)
[2026-04-13 22:57:06,457] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect)
[2026-04-13 22:57:06,461] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect)
[2026-04-13 22:57:06,468] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect)
[2026-04-13 22:57:06,480] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect)
[2026-04-13 22:57:06,504] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect)
[2026-04-13 22:57:06,506] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect)
[2026-04-13 22:57:06,510] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect)
[2026-04-13 22:57:06,519] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect)
[2026-04-13 22:57:13,815] [INFO] [comm.py:675:init_distributed] cdb=None
[2026-04-13 22:57:13,817] [INFO] [comm.py:675:init_distributed] cdb=None
[2026-04-13 22:57:14,049] [INFO] [comm.py:675:init_distributed] cdb=None
[2026-04-13 22:57:14,097] [INFO] [comm.py:675:init_distributed] cdb=None
[2026-04-13 22:57:14,123] [INFO] [comm.py:675:init_distributed] cdb=None
[2026-04-13 22:57:14,212] [INFO] [comm.py:675:init_distributed] cdb=None
[2026-04-13 22:57:14,298] [INFO] [comm.py:675:init_distributed] cdb=None
[2026-04-13 22:57:14,326] [INFO] [comm.py:675:init_distributed] cdb=None
[2026-04-13 22:57:14,331] [INFO] [comm.py:675:init_distributed] cdb=None
[2026-04-13 22:57:14,339] [INFO] [comm.py:675:init_distributed] cdb=None
[2026-04-13 22:57:14,424] [INFO] [comm.py:675:init_distributed] cdb=None
[2026-04-13 22:57:14,502] [INFO] [comm.py:675:init_distributed] cdb=None
[2026-04-13 22:57:14,508] [INFO] [comm.py:675:init_distributed] cdb=None
[2026-04-13 22:57:14,571] [INFO] [comm.py:675:init_distributed] cdb=None
[2026-04-13 22:57:14,570] [INFO] [comm.py:675:init_distributed] cdb=None
[2026-04-13 22:57:14,576] [INFO] [comm.py:675:init_distributed] cdb=None
[2026-04-13 22:57:14,581] [INFO] [comm.py:675:init_distributed] cdb=None
[2026-04-13 22:57:14,593] [INFO] [comm.py:675:init_distributed] cdb=None
[2026-04-13 22:57:14,603] [INFO] [comm.py:675:init_distributed] cdb=None
[2026-04-13 22:57:14,603] [INFO] [comm.py:675:init_distributed] cdb=None
[2026-04-13 22:57:14,635] [INFO] [comm.py:675:init_distributed] cdb=None
[2026-04-13 22:57:14,657] [INFO] [comm.py:675:init_distributed] cdb=None
[2026-04-13 22:57:14,659] [INFO] [comm.py:675:init_distributed] cdb=None
[2026-04-13 22:57:14,667] [INFO] [comm.py:675:init_distributed] cdb=None
[2026-04-13 22:57:14,690] [INFO] [comm.py:675:init_distributed] cdb=None
[2026-04-13 22:57:14,754] [INFO] [comm.py:675:init_distributed] cdb=None
[2026-04-13 22:57:14,768] [INFO] [comm.py:675:init_distributed] cdb=None
[2026-04-13 22:57:14,800] [INFO] [comm.py:675:init_distributed] cdb=None
[2026-04-13 22:57:14,820] [INFO] [comm.py:675:init_distributed] cdb=None
[2026-04-13 22:57:14,894] [INFO] [comm.py:675:init_distributed] cdb=None
[2026-04-13 22:57:15,006] [INFO] [comm.py:675:init_distributed] cdb=None
[2026-04-13 22:57:15,054] [INFO] [comm.py:675:init_distributed] cdb=None
[2026-04-13 22:57:15,358] [INFO] [comm.py:675:init_distributed] cdb=None
[2026-04-13 22:57:15,377] [INFO] [comm.py:675:init_distributed] cdb=None
[2026-04-13 22:57:15,407] [INFO] [comm.py:675:init_distributed] cdb=None
[2026-04-13 22:57:15,465] [INFO] [comm.py:675:init_distributed] cdb=None
[2026-04-13 22:57:15,467] [INFO] [comm.py:675:init_distributed] cdb=None
[2026-04-13 22:57:15,467] [INFO] [comm.py:706:init_distributed] Initializing TorchBackend in DeepSpeed with backend nccl
[2026-04-13 22:57:15,484] [INFO] [comm.py:675:init_distributed] cdb=None
[2026-04-13 22:57:15,485] [INFO] [comm.py:675:init_distributed] cdb=None
[2026-04-13 22:57:15,501] [INFO] [comm.py:675:init_distributed] cdb=None
[2026-04-13 22:57:15,805] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64
[2026-04-13 22:57:15,845] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64
[2026-04-13 22:57:15,845] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64
[2026-04-13 22:57:15,854] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64
[2026-04-13 22:57:15,906] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64
[2026-04-13 22:57:15,910] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64
[2026-04-13 22:57:15,950] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64
[2026-04-13 22:57:15,955] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64
[2026-04-13 22:57:15,979] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64
[2026-04-13 22:57:15,979] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64
[2026-04-13 22:57:15,981] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64
[2026-04-13 22:57:15,981] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64
[2026-04-13 22:57:15,986] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64
[2026-04-13 22:57:15,989] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64
[2026-04-13 22:57:15,992] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64
[2026-04-13 22:57:16,003] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64
[2026-04-13 22:57:16,004] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64
[2026-04-13 22:57:16,009] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64
[2026-04-13 22:57:16,023] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64
[2026-04-13 22:57:16,044] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64
[2026-04-13 22:57:16,050] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64
[2026-04-13 22:57:16,054] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64
[2026-04-13 22:57:22,654] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64
[2026-04-13 22:57:22,662] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64
[2026-04-13 22:57:22,662] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64
[2026-04-13 22:57:22,677] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64
[2026-04-13 22:57:22,680] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64
[2026-04-13 22:57:22,683] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64
[2026-04-13 22:57:22,684] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64
[2026-04-13 22:57:22,684] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64
[2026-04-13 22:57:22,690] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64
[2026-04-13 22:57:22,694] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64
[2026-04-13 22:57:22,710] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64
[2026-04-13 22:57:22,750] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64
[2026-04-13 22:57:22,751] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64
[2026-04-13 22:57:22,753] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64
[2026-04-13 22:57:22,757] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64
[2026-04-13 22:57:22,769] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64
[2026-04-13 22:57:22,775] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64
[2026-04-13 22:57:25,930] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect)
[2026-04-13 22:57:25,930] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect)
[2026-04-13 22:57:25,930] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect)
[2026-04-13 22:57:25,930] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect)
[2026-04-13 22:57:25,930] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect)
[2026-04-13 22:57:25,930] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect)
[2026-04-13 22:57:25,930] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect)
[2026-04-13 22:57:25,930] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect)
[2026-04-13 22:57:26,065] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect)
[2026-04-13 22:57:26,065] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect)
[2026-04-13 22:57:26,065] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect)
[2026-04-13 22:57:26,065] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect)
[2026-04-13 22:57:26,065] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect)
[2026-04-13 22:57:26,065] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect)
[2026-04-13 22:57:26,065] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect)
[2026-04-13 22:57:26,065] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect)
[2026-04-13 22:57:26,534] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect)
[2026-04-13 22:57:26,534] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect)
[2026-04-13 22:57:26,534] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect)
[2026-04-13 22:57:26,534] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect)
[2026-04-13 22:57:26,534] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect)
[2026-04-13 22:57:26,534] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect)
[2026-04-13 22:57:26,534] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect)
[2026-04-13 22:57:26,535] [INFO] [real_accelerator.py:254:get_accelerator] Setting ds_accelerator to cuda (auto detect)
[2026-04-13 22:57:26,700] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64
[2026-04-13 22:57:37,768] [INFO] [comm.py:675:init_distributed] cdb=None
[2026-04-13 22:57:37,768] [INFO] [comm.py:675:init_distributed] cdb=None
[2026-04-13 22:57:37,768] [INFO] [comm.py:675:init_distributed] cdb=None
[2026-04-13 22:57:37,768] [INFO] [comm.py:675:init_distributed] cdb=None
[2026-04-13 22:57:37,768] [INFO] [comm.py:675:init_distributed] cdb=None
[2026-04-13 22:57:37,768] [INFO] [comm.py:675:init_distributed] cdb=None
[2026-04-13 22:57:37,768] [INFO] [comm.py:675:init_distributed] cdb=None
[2026-04-13 22:57:37,769] [INFO] [comm.py:675:init_distributed] cdb=None
[2026-04-13 22:57:37,796] [INFO] [comm.py:675:init_distributed] cdb=None
[2026-04-13 22:57:37,796] [INFO] [comm.py:675:init_distributed] cdb=None
[2026-04-13 22:57:37,796] [INFO] [comm.py:675:init_distributed] cdb=None
[2026-04-13 22:57:37,796] [INFO] [comm.py:675:init_distributed] cdb=None
[2026-04-13 22:57:37,796] [INFO] [comm.py:675:init_distributed] cdb=None
[2026-04-13 22:57:37,796] [INFO] [comm.py:675:init_distributed] cdb=None
[2026-04-13 22:57:37,796] [INFO] [comm.py:675:init_distributed] cdb=None
[2026-04-13 22:57:37,796] [INFO] [comm.py:675:init_distributed] cdb=None
[2026-04-13 22:57:38,359] [INFO] [comm.py:675:init_distributed] cdb=None
[2026-04-13 22:57:38,359] [INFO] [comm.py:675:init_distributed] cdb=None
[2026-04-13 22:57:38,359] [INFO] [comm.py:675:init_distributed] cdb=None
[2026-04-13 22:57:38,359] [INFO] [comm.py:675:init_distributed] cdb=None
[2026-04-13 22:57:38,359] [INFO] [comm.py:675:init_distributed] cdb=None
[2026-04-13 22:57:38,359] [INFO] [comm.py:675:init_distributed] cdb=None
[2026-04-13 22:57:38,359] [INFO] [comm.py:675:init_distributed] cdb=None
[2026-04-13 22:57:38,360] [INFO] [comm.py:675:init_distributed] cdb=None
[2026-04-13 22:57:40,194] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64
[2026-04-13 22:57:40,194] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64
[2026-04-13 22:57:40,194] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64
[2026-04-13 22:57:40,194] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64
[2026-04-13 22:57:40,194] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64
[2026-04-13 22:57:40,215] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64
[2026-04-13 22:57:40,215] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64
[2026-04-13 22:57:40,215] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64
[2026-04-13 22:57:40,215] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64
[2026-04-13 22:57:40,215] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64
[2026-04-13 22:57:40,215] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64
[2026-04-13 22:57:40,216] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64
[2026-04-13 22:57:40,363] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64
[2026-04-13 22:57:40,363] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64
[2026-04-13 22:57:40,363] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64
[2026-04-13 22:57:40,363] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64
[2026-04-13 22:57:40,364] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64
[2026-04-13 22:57:47,001] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64
[2026-04-13 22:57:47,049] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64
[2026-04-13 22:57:47,061] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64
[2026-04-13 22:57:47,091] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64
[2026-04-13 22:57:47,098] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64
[2026-04-13 22:57:47,098] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64
[2026-04-13 22:57:47,105] [INFO] [config.py:744:__init__] Config mesh_device None world_size = 64
[2026-04-13 22:57:50,876] [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-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-64Using 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-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 Prefix Tree collator
args.report_to: ['wandb']
args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-57-14_batch-block1-1089
Using Prefix Tree collator
args.report_to: ['wandb']
args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-57-14_batch-block1-3035
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_22-57-14_batch-block1-1062
args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-57-14_batch-block1-1089
args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-57-13_batch-block1-3544
Using Prefix Tree collator
Using Prefix Tree collator
args.report_to: ['wandb']
Using Prefix Tree collator
args.report_to: ['wandb']
args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-57-14_batch-block1-1062
args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-57-15_batch-block1-3736
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_22-57-14_batch-block1-3544
Using Prefix Tree collator
args.report_to: ['wandb']
args.report_to: ['wandb']
args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-57-15_batch-block1-3736
args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-57-13_batch-block1-3544
Using Prefix Tree collator
args.report_to: ['wandb']
args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-57-15_batch-block1-3736
args.report_to: ['wandb']
Using Prefix Tree collator
args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-57-14_batch-block1-3035
args.report_to: ['wandb']
Using Prefix Tree collator
args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-57-15_batch-block1-3736
args.report_to: ['wandb']
Using Prefix Tree collator
Using Prefix Tree collator
Using Prefix Tree collator
args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-57-14_batch-block1-1062
Using Prefix Tree collator
args.report_to: ['wandb']
Using Prefix Tree collator
args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-57-14_batch-block1-1089
args.report_to: ['wandb']
args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-57-14_batch-block1-1089
args.report_to: ['wandb']
args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-57-14_batch-block1-1089
Using Prefix Tree collator
args.report_to: ['wandb']
args.report_to: ['wandb']
args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-57-14_batch-block1-1062
args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-57-14_batch-block1-3544
args.report_to: ['wandb']
args.report_to: ['wandb']
args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-57-14_batch-block1-3544
args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-57-14_batch-block1-1062
Using Prefix Tree collator
Using Prefix Tree collator
args.report_to: ['wandb']
args.report_to: ['wandb']
args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-57-15_batch-block1-1062
args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-57-15_batch-block1-3736
Using Prefix Tree collator
args.report_to: ['wandb']
args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-57-14_batch-block1-3035
Using Prefix Tree collator
args.report_to: ['wandb']
args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-57-14_batch-block1-1089
Using Prefix Tree collator
Using Prefix Tree collator
Using Prefix Tree collator
args.report_to: ['wandb']
args.report_to: ['wandb']
Using Prefix Tree collator
Using Prefix Tree collator
args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-57-14_batch-block1-1062
args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-57-14_batch-block1-3544
Using Prefix Tree collator
args.report_to: ['wandb']Using Prefix Tree collator
args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-57-14_batch-block1-3035
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_22-57-14_batch-block1-3035
args.report_to: ['wandb']
args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-57-14_batch-block1-3035
args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-57-14_batch-block1-1089
args.report_to: ['wandb']
args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-57-15_batch-block1-3736
args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-57-14_batch-block1-1089
Using Prefix Tree collator
args.report_to: ['wandb']
args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-57-14_batch-block1-3544
Using Prefix Tree collator
Using Prefix Tree collator
args.report_to: ['wandb']
args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-57-14_batch-block1-3544
args.report_to: ['wandb']
Using Prefix Tree collatorUsing Prefix Tree collator
args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-57-15_batch-block1-1062
args.report_to: ['wandb']
args.report_to: ['wandb']args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-57-14_batch-block1-3035
args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-57-14_batch-block1-3035
Using Prefix Tree collator
Using Prefix Tree collator
args.report_to: ['wandb']
args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-57-15_batch-block1-3736
args.report_to: ['wandb']
args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-57-15_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
args.report_to: ['wandb']
args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-57-37_batch-block1-3163
args.report_to: ['wandb']
args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-57-37_batch-block1-3163
args.report_to: ['wandb']
args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-57-37_batch-block1-3163
args.report_to: ['wandb']
args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-57-37_batch-block1-3163args.report_to: ['wandb']
args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-57-37_batch-block1-3163
args.report_to: ['wandb']
args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-57-37_batch-block1-3163
Using Prefix Tree collator
args.report_to: ['wandb']
args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-57-37_batch-block1-3163
args.report_to: ['wandb']
args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-57-37_batch-block1-3163
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 collator
Using Prefix Tree collator
args.report_to: ['wandb']
args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-57-38_batch-block1-3305
args.report_to: ['wandb']
args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-57-38_batch-block1-3305
args.report_to: ['wandb']
args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-57-38_batch-block1-3305args.report_to: ['wandb']
args.report_to: ['wandb']
args.report_to: ['wandb']args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-57-38_batch-block1-3305
args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-57-38_batch-block1-3305
args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-57-38_batch-block1-3305
args.report_to: ['wandb']
args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-57-38_batch-block1-3305
args.report_to: ['wandb']
args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-57-38_batch-block1-3305
Using Prefix Tree collator
Using Prefix Tree collator
Using Prefix Tree collator
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_22-57-37_batch-block1-3038
args.report_to: ['wandb']
args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-57-37_batch-block1-3038
args.report_to: ['wandb']
args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-57-37_batch-block1-3038
args.report_to: ['wandb']
args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-57-37_batch-block1-3038
args.report_to: ['wandb']
args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-57-37_batch-block1-3038
args.report_to: ['wandb']args.report_to: ['wandb']
args.report_to: ['wandb']args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-57-37_batch-block1-3038
args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-57-37_batch-block1-3038
args.logging_dir: runs/dev/tw-8node-resume/runs/Apr13_22-57-37_batch-block1-3038
Parameter Offload: Total persistent parameters: 308224 in 145 params
{'train_runtime': 4.1282, 'train_samples_per_second': 1868.144, 'train_steps_per_second': 15.503, '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