[2025-05-26 14:41:08,996] [INFO] [real_accelerator.py:239:get_accelerator] Setting ds_accelerator to cuda (auto detect)
Warning: The cache directory for DeepSpeed Triton autotune, /home/hansirui_1st/.triton/autotune, appears to be on an NFS system. While this is generally acceptable, if you experience slowdowns or hanging when DeepSpeed exits, it is recommended to set the TRITON_CACHE_DIR environment variable to a non-NFS path.
[2025-05-26 14:41:14,075] [WARNING] [runner.py:215:fetch_hostfile] Unable to find hostfile, will proceed with training with local resources only.
[2025-05-26 14:41:14,075] [INFO] [runner.py:605:main] cmd = /aifs4su/hansirui_1st/miniconda3/envs/jy-resist/bin/python3.11 -u -m deepspeed.launcher.launch --world_info=eyJsb2NhbGhvc3QiOiBbMCwgMSwgMiwgMywgNCwgNSwgNiwgN119 --master_addr=127.0.0.1 --master_port=49734 --module --enable_each_rank_log=None safe_rlhf.finetune --train_datasets inverse-json::/home/hansirui_1st/jiayi/resist/imdb_data/train/pos/5000/train.json --model_name_or_path /aifs4su/hansirui_1st/models/Qwen1.5-0.5B --max_length 512 --trust_remote_code True --epochs 1 --per_device_train_batch_size 1 --per_device_eval_batch_size 4 --gradient_accumulation_steps 8 --gradient_checkpointing --learning_rate 1e-5 --lr_warmup_ratio 0 --weight_decay 0.0 --lr_scheduler_type constant --weight_decay 0.0 --seed 42 --output_dir /aifs4su/hansirui_1st/jiayi/setting3-imdb/Qwen1.5-0.5B/Qwen1.5-0.5B-s3-Q1-5000 --log_type wandb --log_run_name imdb-Qwen1.5-0.5B-s3-Q1-5000 --log_project Inverse_Alignment_IMDb --zero_stage 3 --offload none --bf16 True --tf32 True --save_16bit
[2025-05-26 14:41:16,342] [INFO] [real_accelerator.py:239:get_accelerator] Setting ds_accelerator to cuda (auto detect)
Warning: The cache directory for DeepSpeed Triton autotune, /home/hansirui_1st/.triton/autotune, appears to be on an NFS system. While this is generally acceptable, if you experience slowdowns or hanging when DeepSpeed exits, it is recommended to set the TRITON_CACHE_DIR environment variable to a non-NFS path.
[2025-05-26 14:41:21,317] [INFO] [launch.py:146:main] WORLD INFO DICT: {'localhost': [0, 1, 2, 3, 4, 5, 6, 7]}
[2025-05-26 14:41:21,318] [INFO] [launch.py:152:main] nnodes=1, num_local_procs=8, node_rank=0
[2025-05-26 14:41:21,318] [INFO] [launch.py:163:main] global_rank_mapping=defaultdict(<class 'list'>, {'localhost': [0, 1, 2, 3, 4, 5, 6, 7]})
[2025-05-26 14:41:21,318] [INFO] [launch.py:164:main] dist_world_size=8
[2025-05-26 14:41:21,318] [INFO] [launch.py:168:main] Setting CUDA_VISIBLE_DEVICES=0,1,2,3,4,5,6,7
[2025-05-26 14:41:21,318] [INFO] [launch.py:256:main] process 3396231 spawned with command: ['/aifs4su/hansirui_1st/miniconda3/envs/jy-resist/bin/python3.11', '-u', '-m', 'safe_rlhf.finetune', '--local_rank=0', '--train_datasets', 'inverse-json::/home/hansirui_1st/jiayi/resist/imdb_data/train/pos/5000/train.json', '--model_name_or_path', '/aifs4su/hansirui_1st/models/Qwen1.5-0.5B', '--max_length', '512', '--trust_remote_code', 'True', '--epochs', '1', '--per_device_train_batch_size', '1', '--per_device_eval_batch_size', '4', '--gradient_accumulation_steps', '8', '--gradient_checkpointing', '--learning_rate', '1e-5', '--lr_warmup_ratio', '0', '--weight_decay', '0.0', '--lr_scheduler_type', 'constant', '--weight_decay', '0.0', '--seed', '42', '--output_dir', '/aifs4su/hansirui_1st/jiayi/setting3-imdb/Qwen1.5-0.5B/Qwen1.5-0.5B-s3-Q1-5000', '--log_type', 'wandb', '--log_run_name', 'imdb-Qwen1.5-0.5B-s3-Q1-5000', '--log_project', 'Inverse_Alignment_IMDb', '--zero_stage', '3', '--offload', 'none', '--bf16', 'True', '--tf32', 'True', '--save_16bit']
[2025-05-26 14:41:21,319] [INFO] [launch.py:256:main] process 3396232 spawned with command: ['/aifs4su/hansirui_1st/miniconda3/envs/jy-resist/bin/python3.11', '-u', '-m', 'safe_rlhf.finetune', '--local_rank=1', '--train_datasets', 'inverse-json::/home/hansirui_1st/jiayi/resist/imdb_data/train/pos/5000/train.json', '--model_name_or_path', '/aifs4su/hansirui_1st/models/Qwen1.5-0.5B', '--max_length', '512', '--trust_remote_code', 'True', '--epochs', '1', '--per_device_train_batch_size', '1', '--per_device_eval_batch_size', '4', '--gradient_accumulation_steps', '8', '--gradient_checkpointing', '--learning_rate', '1e-5', '--lr_warmup_ratio', '0', '--weight_decay', '0.0', '--lr_scheduler_type', 'constant', '--weight_decay', '0.0', '--seed', '42', '--output_dir', '/aifs4su/hansirui_1st/jiayi/setting3-imdb/Qwen1.5-0.5B/Qwen1.5-0.5B-s3-Q1-5000', '--log_type', 'wandb', '--log_run_name', 'imdb-Qwen1.5-0.5B-s3-Q1-5000', '--log_project', 'Inverse_Alignment_IMDb', '--zero_stage', '3', '--offload', 'none', '--bf16', 'True', '--tf32', 'True', '--save_16bit']
[2025-05-26 14:41:21,319] [INFO] [launch.py:256:main] process 3396233 spawned with command: ['/aifs4su/hansirui_1st/miniconda3/envs/jy-resist/bin/python3.11', '-u', '-m', 'safe_rlhf.finetune', '--local_rank=2', '--train_datasets', 'inverse-json::/home/hansirui_1st/jiayi/resist/imdb_data/train/pos/5000/train.json', '--model_name_or_path', '/aifs4su/hansirui_1st/models/Qwen1.5-0.5B', '--max_length', '512', '--trust_remote_code', 'True', '--epochs', '1', '--per_device_train_batch_size', '1', '--per_device_eval_batch_size', '4', '--gradient_accumulation_steps', '8', '--gradient_checkpointing', '--learning_rate', '1e-5', '--lr_warmup_ratio', '0', '--weight_decay', '0.0', '--lr_scheduler_type', 'constant', '--weight_decay', '0.0', '--seed', '42', '--output_dir', '/aifs4su/hansirui_1st/jiayi/setting3-imdb/Qwen1.5-0.5B/Qwen1.5-0.5B-s3-Q1-5000', '--log_type', 'wandb', '--log_run_name', 'imdb-Qwen1.5-0.5B-s3-Q1-5000', '--log_project', 'Inverse_Alignment_IMDb', '--zero_stage', '3', '--offload', 'none', '--bf16', 'True', '--tf32', 'True', '--save_16bit']
[2025-05-26 14:41:21,320] [INFO] [launch.py:256:main] process 3396234 spawned with command: ['/aifs4su/hansirui_1st/miniconda3/envs/jy-resist/bin/python3.11', '-u', '-m', 'safe_rlhf.finetune', '--local_rank=3', '--train_datasets', 'inverse-json::/home/hansirui_1st/jiayi/resist/imdb_data/train/pos/5000/train.json', '--model_name_or_path', '/aifs4su/hansirui_1st/models/Qwen1.5-0.5B', '--max_length', '512', '--trust_remote_code', 'True', '--epochs', '1', '--per_device_train_batch_size', '1', '--per_device_eval_batch_size', '4', '--gradient_accumulation_steps', '8', '--gradient_checkpointing', '--learning_rate', '1e-5', '--lr_warmup_ratio', '0', '--weight_decay', '0.0', '--lr_scheduler_type', 'constant', '--weight_decay', '0.0', '--seed', '42', '--output_dir', '/aifs4su/hansirui_1st/jiayi/setting3-imdb/Qwen1.5-0.5B/Qwen1.5-0.5B-s3-Q1-5000', '--log_type', 'wandb', '--log_run_name', 'imdb-Qwen1.5-0.5B-s3-Q1-5000', '--log_project', 'Inverse_Alignment_IMDb', '--zero_stage', '3', '--offload', 'none', '--bf16', 'True', '--tf32', 'True', '--save_16bit']
[2025-05-26 14:41:21,320] [INFO] [launch.py:256:main] process 3396235 spawned with command: ['/aifs4su/hansirui_1st/miniconda3/envs/jy-resist/bin/python3.11', '-u', '-m', 'safe_rlhf.finetune', '--local_rank=4', '--train_datasets', 'inverse-json::/home/hansirui_1st/jiayi/resist/imdb_data/train/pos/5000/train.json', '--model_name_or_path', '/aifs4su/hansirui_1st/models/Qwen1.5-0.5B', '--max_length', '512', '--trust_remote_code', 'True', '--epochs', '1', '--per_device_train_batch_size', '1', '--per_device_eval_batch_size', '4', '--gradient_accumulation_steps', '8', '--gradient_checkpointing', '--learning_rate', '1e-5', '--lr_warmup_ratio', '0', '--weight_decay', '0.0', '--lr_scheduler_type', 'constant', '--weight_decay', '0.0', '--seed', '42', '--output_dir', '/aifs4su/hansirui_1st/jiayi/setting3-imdb/Qwen1.5-0.5B/Qwen1.5-0.5B-s3-Q1-5000', '--log_type', 'wandb', '--log_run_name', 'imdb-Qwen1.5-0.5B-s3-Q1-5000', '--log_project', 'Inverse_Alignment_IMDb', '--zero_stage', '3', '--offload', 'none', '--bf16', 'True', '--tf32', 'True', '--save_16bit']
[2025-05-26 14:41:21,321] [INFO] [launch.py:256:main] process 3396236 spawned with command: ['/aifs4su/hansirui_1st/miniconda3/envs/jy-resist/bin/python3.11', '-u', '-m', 'safe_rlhf.finetune', '--local_rank=5', '--train_datasets', 'inverse-json::/home/hansirui_1st/jiayi/resist/imdb_data/train/pos/5000/train.json', '--model_name_or_path', '/aifs4su/hansirui_1st/models/Qwen1.5-0.5B', '--max_length', '512', '--trust_remote_code', 'True', '--epochs', '1', '--per_device_train_batch_size', '1', '--per_device_eval_batch_size', '4', '--gradient_accumulation_steps', '8', '--gradient_checkpointing', '--learning_rate', '1e-5', '--lr_warmup_ratio', '0', '--weight_decay', '0.0', '--lr_scheduler_type', 'constant', '--weight_decay', '0.0', '--seed', '42', '--output_dir', '/aifs4su/hansirui_1st/jiayi/setting3-imdb/Qwen1.5-0.5B/Qwen1.5-0.5B-s3-Q1-5000', '--log_type', 'wandb', '--log_run_name', 'imdb-Qwen1.5-0.5B-s3-Q1-5000', '--log_project', 'Inverse_Alignment_IMDb', '--zero_stage', '3', '--offload', 'none', '--bf16', 'True', '--tf32', 'True', '--save_16bit']
[2025-05-26 14:41:21,321] [INFO] [launch.py:256:main] process 3396237 spawned with command: ['/aifs4su/hansirui_1st/miniconda3/envs/jy-resist/bin/python3.11', '-u', '-m', 'safe_rlhf.finetune', '--local_rank=6', '--train_datasets', 'inverse-json::/home/hansirui_1st/jiayi/resist/imdb_data/train/pos/5000/train.json', '--model_name_or_path', '/aifs4su/hansirui_1st/models/Qwen1.5-0.5B', '--max_length', '512', '--trust_remote_code', 'True', '--epochs', '1', '--per_device_train_batch_size', '1', '--per_device_eval_batch_size', '4', '--gradient_accumulation_steps', '8', '--gradient_checkpointing', '--learning_rate', '1e-5', '--lr_warmup_ratio', '0', '--weight_decay', '0.0', '--lr_scheduler_type', 'constant', '--weight_decay', '0.0', '--seed', '42', '--output_dir', '/aifs4su/hansirui_1st/jiayi/setting3-imdb/Qwen1.5-0.5B/Qwen1.5-0.5B-s3-Q1-5000', '--log_type', 'wandb', '--log_run_name', 'imdb-Qwen1.5-0.5B-s3-Q1-5000', '--log_project', 'Inverse_Alignment_IMDb', '--zero_stage', '3', '--offload', 'none', '--bf16', 'True', '--tf32', 'True', '--save_16bit']
[2025-05-26 14:41:21,322] [INFO] [launch.py:256:main] process 3396238 spawned with command: ['/aifs4su/hansirui_1st/miniconda3/envs/jy-resist/bin/python3.11', '-u', '-m', 'safe_rlhf.finetune', '--local_rank=7', '--train_datasets', 'inverse-json::/home/hansirui_1st/jiayi/resist/imdb_data/train/pos/5000/train.json', '--model_name_or_path', '/aifs4su/hansirui_1st/models/Qwen1.5-0.5B', '--max_length', '512', '--trust_remote_code', 'True', '--epochs', '1', '--per_device_train_batch_size', '1', '--per_device_eval_batch_size', '4', '--gradient_accumulation_steps', '8', '--gradient_checkpointing', '--learning_rate', '1e-5', '--lr_warmup_ratio', '0', '--weight_decay', '0.0', '--lr_scheduler_type', 'constant', '--weight_decay', '0.0', '--seed', '42', '--output_dir', '/aifs4su/hansirui_1st/jiayi/setting3-imdb/Qwen1.5-0.5B/Qwen1.5-0.5B-s3-Q1-5000', '--log_type', 'wandb', '--log_run_name', 'imdb-Qwen1.5-0.5B-s3-Q1-5000', '--log_project', 'Inverse_Alignment_IMDb', '--zero_stage', '3', '--offload', 'none', '--bf16', 'True', '--tf32', 'True', '--save_16bit']
[2025-05-26 14:41:25,577] [INFO] [real_accelerator.py:239:get_accelerator] Setting ds_accelerator to cuda (auto detect)
Warning: The cache directory for DeepSpeed Triton autotune, /home/hansirui_1st/.triton/autotune, appears to be on an NFS system. While this is generally acceptable, if you experience slowdowns or hanging when DeepSpeed exits, it is recommended to set the TRITON_CACHE_DIR environment variable to a non-NFS path.
[2025-05-26 14:41:25,898] [INFO] [real_accelerator.py:239:get_accelerator] Setting ds_accelerator to cuda (auto detect)
[2025-05-26 14:41:26,004] [INFO] [real_accelerator.py:239:get_accelerator] Setting ds_accelerator to cuda (auto detect)
[2025-05-26 14:41:26,038] [INFO] [real_accelerator.py:239:get_accelerator] Setting ds_accelerator to cuda (auto detect)
Warning: The cache directory for DeepSpeed Triton autotune, /home/hansirui_1st/.triton/autotune, appears to be on an NFS system. While this is generally acceptable, if you experience slowdowns or hanging when DeepSpeed exits, it is recommended to set the TRITON_CACHE_DIR environment variable to a non-NFS path.
[2025-05-26 14:41:26,130] [INFO] [real_accelerator.py:239:get_accelerator] Setting ds_accelerator to cuda (auto detect)
[2025-05-26 14:41:26,140] [INFO] [real_accelerator.py:239:get_accelerator] Setting ds_accelerator to cuda (auto detect)
[2025-05-26 14:41:26,188] [INFO] [real_accelerator.py:239:get_accelerator] Setting ds_accelerator to cuda (auto detect)
[2025-05-26 14:41:26,206] [INFO] [real_accelerator.py:239:get_accelerator] Setting ds_accelerator to cuda (auto detect)
Warning: The cache directory for DeepSpeed Triton autotune, /home/hansirui_1st/.triton/autotune, appears to be on an NFS system. While this is generally acceptable, if you experience slowdowns or hanging when DeepSpeed exits, it is recommended to set the TRITON_CACHE_DIR environment variable to a non-NFS path.
Warning: The cache directory for DeepSpeed Triton autotune, /home/hansirui_1st/.triton/autotune, appears to be on an NFS system. While this is generally acceptable, if you experience slowdowns or hanging when DeepSpeed exits, it is recommended to set the TRITON_CACHE_DIR environment variable to a non-NFS path.
Warning: The cache directory for DeepSpeed Triton autotune, /home/hansirui_1st/.triton/autotune, appears to be on an NFS system. While this is generally acceptable, if you experience slowdowns or hanging when DeepSpeed exits, it is recommended to set the TRITON_CACHE_DIR environment variable to a non-NFS path.
Warning: The cache directory for DeepSpeed Triton autotune, /home/hansirui_1st/.triton/autotune, appears to be on an NFS system. While this is generally acceptable, if you experience slowdowns or hanging when DeepSpeed exits, it is recommended to set the TRITON_CACHE_DIR environment variable to a non-NFS path.
Warning: The cache directory for DeepSpeed Triton autotune, /home/hansirui_1st/.triton/autotune, appears to be on an NFS system. While this is generally acceptable, if you experience slowdowns or hanging when DeepSpeed exits, it is recommended to set the TRITON_CACHE_DIR environment variable to a non-NFS path.
Warning: The cache directory for DeepSpeed Triton autotune, /home/hansirui_1st/.triton/autotune, appears to be on an NFS system. While this is generally acceptable, if you experience slowdowns or hanging when DeepSpeed exits, it is recommended to set the TRITON_CACHE_DIR environment variable to a non-NFS path.
[2025-05-26 14:41:33,028] [INFO] [comm.py:669:init_distributed] cdb=None
[2025-05-26 14:41:33,028] [INFO] [comm.py:700:init_distributed] Initializing TorchBackend in DeepSpeed with backend nccl
[2025-05-26 14:41:33,350] [INFO] [comm.py:669:init_distributed] cdb=None
[2025-05-26 14:41:33,436] [INFO] [comm.py:669:init_distributed] cdb=None
[2025-05-26 14:41:33,532] [INFO] [comm.py:669:init_distributed] cdb=None
[2025-05-26 14:41:33,584] [INFO] [comm.py:669:init_distributed] cdb=None
[2025-05-26 14:41:33,624] [INFO] [comm.py:669:init_distributed] cdb=None
[2025-05-26 14:41:33,761] [INFO] [comm.py:669:init_distributed] cdb=None
[2025-05-26 14:41:33,770] [INFO] [comm.py:669:init_distributed] cdb=None
Set logger level to INFO.
[2025-05-26 14:41:44,584] [INFO] [config.py:735:__init__] Config mesh_device None world_size = 8
[2025-05-26 14:41:44,584] [INFO] [config.py:735:__init__] Config mesh_device None world_size = 8
[2025-05-26 14:41:44,584] [INFO] [config.py:735:__init__] Config mesh_device None world_size = 8
[2025-05-26 14:41:44,584] [INFO] [config.py:735:__init__] Config mesh_device None world_size = 8
[2025-05-26 14:41:44,584] [INFO] [config.py:735:__init__] Config mesh_device None world_size = 8
[2025-05-26 14:41:44,584] [INFO] [config.py:735:__init__] Config mesh_device None world_size = 8
[2025-05-26 14:41:44,585] [INFO] [config.py:735:__init__] Config mesh_device None world_size = 8
[2025-05-26 14:41:44,590] [INFO] [config.py:735:__init__] Config mesh_device None world_size = 8
[2025-05-26 14:41:45,032] [INFO] [partition_parameters.py:348:__exit__] finished initializing model - num_params = 291, num_elems = 0.62B
ninja: no work to do.
Time to load fused_adam op: 0.1025545597076416 seconds
Time to load fused_adam op: 0.10345292091369629 secondsTime to load fused_adam op: 0.10343718528747559 seconds

Time to load fused_adam op: 0.10253119468688965 seconds
Time to load fused_adam op: 0.10353612899780273 seconds
Time to load fused_adam op: 0.10224509239196777 seconds
[2025-05-26 14:41:46,961] [INFO] [config.py:735:__init__] Config mesh_device None world_size = 8
Time to load fused_adam op: 0.1016840934753418 seconds
[2025-05-26 14:41:46,963] [INFO] [config.py:735:__init__] Config mesh_device None world_size = 8
[2025-05-26 14:41:46,963] [INFO] [config.py:735:__init__] Config mesh_device None world_size = 8
[2025-05-26 14:41:46,964] [INFO] [logging.py:107:log_dist] [Rank 0] DeepSpeed info: version=0.16.8, git-hash=unknown, git-branch=unknown
[2025-05-26 14:41:46,964] [INFO] [config.py:735:__init__] Config mesh_device None world_size = 8
[2025-05-26 14:41:46,964] [INFO] [config.py:735:__init__] Config mesh_device None world_size = 8
[2025-05-26 14:41:46,964] [INFO] [comm.py:694:init_distributed] Distributed backend already initialized
[2025-05-26 14:41:46,964] [INFO] [config.py:735:__init__] Config mesh_device None world_size = 8
[2025-05-26 14:41:46,964] [INFO] [config.py:735:__init__] Config mesh_device None world_size = 8
Time to load fused_adam op: 0.1022486686706543 seconds
[2025-05-26 14:41:46,966] [INFO] [config.py:735:__init__] Config mesh_device None world_size = 8
[2025-05-26 14:41:46,968] [INFO] [logging.py:107:log_dist] [Rank 0] DeepSpeed Flops Profiler Enabled: False
[2025-05-26 14:41:46,969] [INFO] [logging.py:107:log_dist] [Rank 0] Using client Optimizer as basic optimizer
[2025-05-26 14:41:46,969] [INFO] [logging.py:107:log_dist] [Rank 0] Removing param_group that has no 'params' in the basic Optimizer
[2025-05-26 14:41:46,973] [INFO] [logging.py:107:log_dist] [Rank 0] DeepSpeed Basic Optimizer = FusedAdam
[2025-05-26 14:41:46,973] [INFO] [utils.py:59:is_zero_supported_optimizer] Checking ZeRO support for optimizer=FusedAdam type=<class 'deepspeed.ops.adam.fused_adam.FusedAdam'>
[2025-05-26 14:41:46,973] [INFO] [logging.py:107:log_dist] [Rank 0] Creating fp16 ZeRO stage 3 optimizer, MiCS is enabled False, Hierarchical params gather False
[2025-05-26 14:41:46,973] [INFO] [logging.py:107:log_dist] [Rank 0] Creating torch.bfloat16 ZeRO stage 3 optimizer
[2025-05-26 14:41:47,150] [INFO] [utils.py:781:see_memory_usage] Stage 3 initialize beginning
[2025-05-26 14:41:47,151] [INFO] [utils.py:782:see_memory_usage] MA 0.36 GB         Max_MA 0.98 GB         CA 1.32 GB         Max_CA 1 GB 
[2025-05-26 14:41:47,151] [INFO] [utils.py:789:see_memory_usage] CPU Virtual Memory:  used = 192.57 GB, percent = 9.6%
[2025-05-26 14:41:47,152] [INFO] [stage3.py:170:__init__] Reduce bucket size 500000000
[2025-05-26 14:41:47,152] [INFO] [stage3.py:171:__init__] Prefetch bucket size 30000000
[2025-05-26 14:41:47,258] [INFO] [utils.py:781:see_memory_usage] DeepSpeedZeRoOffload initialize [begin]
[2025-05-26 14:41:47,258] [INFO] [utils.py:782:see_memory_usage] MA 0.36 GB         Max_MA 0.36 GB         CA 1.32 GB         Max_CA 1 GB 
[2025-05-26 14:41:47,258] [INFO] [utils.py:789:see_memory_usage] CPU Virtual Memory:  used = 192.58 GB, percent = 9.6%
Parameter Offload: Total persistent parameters: 123904 in 121 params
[2025-05-26 14:41:47,392] [INFO] [utils.py:781:see_memory_usage] DeepSpeedZeRoOffload initialize [end]
[2025-05-26 14:41:47,392] [INFO] [utils.py:782:see_memory_usage] MA 0.11 GB         Max_MA 0.4 GB         CA 1.32 GB         Max_CA 1 GB 
[2025-05-26 14:41:47,392] [INFO] [utils.py:789:see_memory_usage] CPU Virtual Memory:  used = 192.58 GB, percent = 9.6%
[2025-05-26 14:41:47,503] [INFO] [utils.py:781:see_memory_usage] Before creating fp16 partitions
[2025-05-26 14:41:47,503] [INFO] [utils.py:782:see_memory_usage] MA 0.11 GB         Max_MA 0.11 GB         CA 1.32 GB         Max_CA 1 GB 
[2025-05-26 14:41:47,503] [INFO] [utils.py:789:see_memory_usage] CPU Virtual Memory:  used = 192.57 GB, percent = 9.6%
[2025-05-26 14:41:47,982] [INFO] [utils.py:781:see_memory_usage] After creating fp16 partitions: 2
[2025-05-26 14:41:47,983] [INFO] [utils.py:782:see_memory_usage] MA 0.11 GB         Max_MA 0.11 GB         CA 0.11 GB         Max_CA 1 GB 
[2025-05-26 14:41:47,983] [INFO] [utils.py:789:see_memory_usage] CPU Virtual Memory:  used = 192.94 GB, percent = 9.6%
[2025-05-26 14:41:48,093] [INFO] [utils.py:781:see_memory_usage] Before creating fp32 partitions
[2025-05-26 14:41:48,094] [INFO] [utils.py:782:see_memory_usage] MA 0.11 GB         Max_MA 0.11 GB         CA 0.11 GB         Max_CA 0 GB 
[2025-05-26 14:41:48,094] [INFO] [utils.py:789:see_memory_usage] CPU Virtual Memory:  used = 192.94 GB, percent = 9.6%
[2025-05-26 14:41:48,211] [INFO] [utils.py:781:see_memory_usage] After creating fp32 partitions
[2025-05-26 14:41:48,211] [INFO] [utils.py:782:see_memory_usage] MA 0.32 GB         Max_MA 0.43 GB         CA 0.44 GB         Max_CA 0 GB 
[2025-05-26 14:41:48,211] [INFO] [utils.py:789:see_memory_usage] CPU Virtual Memory:  used = 192.94 GB, percent = 9.6%
[2025-05-26 14:41:48,325] [INFO] [utils.py:781:see_memory_usage] Before initializing optimizer states
[2025-05-26 14:41:48,325] [INFO] [utils.py:782:see_memory_usage] MA 0.32 GB         Max_MA 0.32 GB         CA 0.44 GB         Max_CA 0 GB 
[2025-05-26 14:41:48,325] [INFO] [utils.py:789:see_memory_usage] CPU Virtual Memory:  used = 192.94 GB, percent = 9.6%
[2025-05-26 14:41:48,438] [INFO] [utils.py:781:see_memory_usage] After initializing optimizer states
[2025-05-26 14:41:48,438] [INFO] [utils.py:782:see_memory_usage] MA 0.32 GB         Max_MA 0.54 GB         CA 0.65 GB         Max_CA 1 GB 
[2025-05-26 14:41:48,438] [INFO] [utils.py:789:see_memory_usage] CPU Virtual Memory:  used = 192.94 GB, percent = 9.6%
[2025-05-26 14:41:48,439] [INFO] [stage3.py:534:_setup_for_real_optimizer] optimizer state initialized
[2025-05-26 14:41:50,005] [INFO] [utils.py:781:see_memory_usage] After initializing ZeRO optimizer
[2025-05-26 14:41:50,005] [INFO] [utils.py:782:see_memory_usage] MA 1.36 GB         Max_MA 1.94 GB         CA 2.17 GB         Max_CA 2 GB 
[2025-05-26 14:41:50,005] [INFO] [utils.py:789:see_memory_usage] CPU Virtual Memory:  used = 192.94 GB, percent = 9.6%
[2025-05-26 14:41:50,005] [INFO] [logging.py:107:log_dist] [Rank 0] DeepSpeed Final Optimizer = DeepSpeedZeroOptimizer_Stage3
[2025-05-26 14:41:50,005] [INFO] [logging.py:107:log_dist] [Rank 0] DeepSpeed using client LR scheduler
[2025-05-26 14:41:50,005] [INFO] [logging.py:107:log_dist] [Rank 0] DeepSpeed LR Scheduler = <torch.optim.lr_scheduler.LambdaLR object at 0x1550e428bc50>
[2025-05-26 14:41:50,005] [INFO] [logging.py:107:log_dist] [Rank 0] step=0, skipped=0, lr=[1e-05, 1e-05], mom=[(0.9, 0.95), (0.9, 0.95)]
[2025-05-26 14:41:50,006] [INFO] [config.py:1003:print] DeepSpeedEngine configuration:
[2025-05-26 14:41:50,006] [INFO] [config.py:1007:print]   activation_checkpointing_config  {
    "partition_activations": false, 
    "contiguous_memory_optimization": false, 
    "cpu_checkpointing": false, 
    "number_checkpoints": null, 
    "synchronize_checkpoint_boundary": false, 
    "profile": false
}
[2025-05-26 14:41:50,006] [INFO] [config.py:1007:print]   aio_config ................... {'block_size': 1048576, 'queue_depth': 8, 'intra_op_parallelism': 1, 'single_submit': False, 'overlap_events': True, 'use_gds': False}
[2025-05-26 14:41:50,006] [INFO] [config.py:1007:print]   amp_enabled .................. False
[2025-05-26 14:41:50,006] [INFO] [config.py:1007:print]   amp_params ................... False
[2025-05-26 14:41:50,006] [INFO] [config.py:1007:print]   autotuning_config ............ {
    "enabled": false, 
    "start_step": null, 
    "end_step": null, 
    "metric_path": null, 
    "arg_mappings": null, 
    "metric": "throughput", 
    "model_info": null, 
    "results_dir": "autotuning_results", 
    "exps_dir": "autotuning_exps", 
    "overwrite": true, 
    "fast": true, 
    "start_profile_step": 3, 
    "end_profile_step": 5, 
    "tuner_type": "gridsearch", 
    "tuner_early_stopping": 5, 
    "tuner_num_trials": 50, 
    "model_info_path": null, 
    "mp_size": 1, 
    "max_train_batch_size": null, 
    "min_train_batch_size": 1, 
    "max_train_micro_batch_size_per_gpu": 1.024000e+03, 
    "min_train_micro_batch_size_per_gpu": 1, 
    "num_tuning_micro_batch_sizes": 3
}
[2025-05-26 14:41:50,006] [INFO] [config.py:1007:print]   bfloat16_enabled ............. True
[2025-05-26 14:41:50,006] [INFO] [config.py:1007:print]   bfloat16_immediate_grad_update  True
[2025-05-26 14:41:50,006] [INFO] [config.py:1007:print]   checkpoint_parallel_write_pipeline  False
[2025-05-26 14:41:50,006] [INFO] [config.py:1007:print]   checkpoint_tag_validation_enabled  True
[2025-05-26 14:41:50,006] [INFO] [config.py:1007:print]   checkpoint_tag_validation_fail  False
[2025-05-26 14:41:50,006] [INFO] [config.py:1007:print]   comms_config ................. <deepspeed.comm.config.DeepSpeedCommsConfig object at 0x1550e4291350>
[2025-05-26 14:41:50,006] [INFO] [config.py:1007:print]   communication_data_type ...... None
[2025-05-26 14:41:50,006] [INFO] [config.py:1007:print]   compile_config ............... deepcompile=False free_activation=False offload_activation=False offload_opt_states=False double_buffer=True symmetric_memory=False debug_log=False offload_parameters=False sync_before_reduce=False sync_after_reduce=False sync_before_allgather=False sync_after_allgather=False
[2025-05-26 14:41:50,006] [INFO] [config.py:1007:print]   compression_config ........... {'weight_quantization': {'shared_parameters': {'enabled': False, 'quantizer_kernel': False, 'schedule_offset': 0, 'quantize_groups': 1, 'quantize_verbose': False, 'quantization_type': 'symmetric', 'quantize_weight_in_forward': False, 'rounding': 'nearest', 'fp16_mixed_quantize': False, 'quantize_change_ratio': 0.001}, 'different_groups': {}}, 'activation_quantization': {'shared_parameters': {'enabled': False, 'quantization_type': 'symmetric', 'range_calibration': 'dynamic', 'schedule_offset': 1000}, 'different_groups': {}}, 'sparse_pruning': {'shared_parameters': {'enabled': False, 'method': 'l1', 'schedule_offset': 1000}, 'different_groups': {}}, 'row_pruning': {'shared_parameters': {'enabled': False, 'method': 'l1', 'schedule_offset': 1000}, 'different_groups': {}}, 'head_pruning': {'shared_parameters': {'enabled': False, 'method': 'topk', 'schedule_offset': 1000}, 'different_groups': {}}, 'channel_pruning': {'shared_parameters': {'enabled': False, 'method': 'l1', 'schedule_offset': 1000}, 'different_groups': {}}, 'layer_reduction': {'enabled': False}}
[2025-05-26 14:41:50,006] [INFO] [config.py:1007:print]   curriculum_enabled_legacy .... False
[2025-05-26 14:41:50,006] [INFO] [config.py:1007:print]   curriculum_params_legacy ..... False
[2025-05-26 14:41:50,006] [INFO] [config.py:1007:print]   data_efficiency_config ....... {'enabled': False, 'seed': 1234, 'data_sampling': {'enabled': False, 'num_epochs': 1000, 'num_workers': 0, 'pin_memory': False, 'curriculum_learning': {'enabled': False}, 'dynamic_batching': {'enabled': False, 'lr_scaling_method': 'linear', 'min_batch_size': 1, 'max_batch_size': None, 'sequence_picking_order': 'dataloader', 'verbose': False}}, 'data_routing': {'enabled': False, 'random_ltd': {'enabled': False, 'layer_token_lr_schedule': {'enabled': False}}}}
[2025-05-26 14:41:50,006] [INFO] [config.py:1007:print]   data_efficiency_enabled ...... False
[2025-05-26 14:41:50,006] [INFO] [config.py:1007:print]   dataloader_drop_last ......... False
[2025-05-26 14:41:50,006] [INFO] [config.py:1007:print]   disable_allgather ............ False
[2025-05-26 14:41:50,006] [INFO] [config.py:1007:print]   dump_state ................... False
[2025-05-26 14:41:50,006] [INFO] [config.py:1007:print]   dynamic_loss_scale_args ...... None
[2025-05-26 14:41:50,007] [INFO] [config.py:1007:print]   eigenvalue_enabled ........... False
[2025-05-26 14:41:50,007] [INFO] [config.py:1007:print]   eigenvalue_gas_boundary_resolution  1
[2025-05-26 14:41:50,007] [INFO] [config.py:1007:print]   eigenvalue_layer_name ........ bert.encoder.layer
[2025-05-26 14:41:50,007] [INFO] [config.py:1007:print]   eigenvalue_layer_num ......... 0
[2025-05-26 14:41:50,007] [INFO] [config.py:1007:print]   eigenvalue_max_iter .......... 100
[2025-05-26 14:41:50,007] [INFO] [config.py:1007:print]   eigenvalue_stability ......... 1e-06
[2025-05-26 14:41:50,007] [INFO] [config.py:1007:print]   eigenvalue_tol ............... 0.01
[2025-05-26 14:41:50,007] [INFO] [config.py:1007:print]   eigenvalue_verbose ........... False
[2025-05-26 14:41:50,007] [INFO] [config.py:1007:print]   elasticity_enabled ........... False
[2025-05-26 14:41:50,007] [INFO] [config.py:1007:print]   flops_profiler_config ........ {
    "enabled": false, 
    "recompute_fwd_factor": 0.0, 
    "profile_step": 1, 
    "module_depth": -1, 
    "top_modules": 1, 
    "detailed": true, 
    "output_file": null
}
[2025-05-26 14:41:50,007] [INFO] [config.py:1007:print]   fp16_auto_cast ............... None
[2025-05-26 14:41:50,007] [INFO] [config.py:1007:print]   fp16_enabled ................. False
[2025-05-26 14:41:50,007] [INFO] [config.py:1007:print]   fp16_master_weights_and_gradients  False
[2025-05-26 14:41:50,007] [INFO] [config.py:1007:print]   global_rank .................. 0
[2025-05-26 14:41:50,007] [INFO] [config.py:1007:print]   grad_accum_dtype ............. None
[2025-05-26 14:41:50,007] [INFO] [config.py:1007:print]   gradient_accumulation_steps .. 8
[2025-05-26 14:41:50,007] [INFO] [config.py:1007:print]   gradient_clipping ............ 1.0
[2025-05-26 14:41:50,007] [INFO] [config.py:1007:print]   gradient_predivide_factor .... 1.0
[2025-05-26 14:41:50,007] [INFO] [config.py:1007:print]   graph_harvesting ............. False
[2025-05-26 14:41:50,007] [INFO] [config.py:1007:print]   hybrid_engine ................ enabled=False max_out_tokens=512 inference_tp_size=1 release_inference_cache=False pin_parameters=True tp_gather_partition_size=8
[2025-05-26 14:41:50,007] [INFO] [config.py:1007:print]   initial_dynamic_scale ........ 1
[2025-05-26 14:41:50,007] [INFO] [config.py:1007:print]   load_universal_checkpoint .... False
[2025-05-26 14:41:50,007] [INFO] [config.py:1007:print]   loss_scale ................... 1.0
[2025-05-26 14:41:50,007] [INFO] [config.py:1007:print]   memory_breakdown ............. False
[2025-05-26 14:41:50,007] [INFO] [config.py:1007:print]   mics_hierarchial_params_gather  False
[2025-05-26 14:41:50,007] [INFO] [config.py:1007:print]   mics_shard_size .............. -1
[2025-05-26 14:41:50,007] [INFO] [config.py:1007:print]   monitor_config ............... tensorboard=TensorBoardConfig(enabled=False, output_path='', job_name='DeepSpeedJobName') comet=CometConfig(enabled=False, samples_log_interval=100, project=None, workspace=None, api_key=None, experiment_name=None, experiment_key=None, online=None, mode=None) wandb=WandbConfig(enabled=False, group=None, team=None, project='deepspeed') csv_monitor=CSVConfig(enabled=False, output_path='', job_name='DeepSpeedJobName')
[2025-05-26 14:41:50,007] [INFO] [config.py:1007:print]   nebula_config ................ {
    "enabled": false, 
    "persistent_storage_path": null, 
    "persistent_time_interval": 100, 
    "num_of_version_in_retention": 2, 
    "enable_nebula_load": true, 
    "load_path": null
}
[2025-05-26 14:41:50,007] [INFO] [config.py:1007:print]   optimizer_legacy_fusion ...... False
[2025-05-26 14:41:50,007] [INFO] [config.py:1007:print]   optimizer_name ............... None
[2025-05-26 14:41:50,007] [INFO] [config.py:1007:print]   optimizer_params ............. None
[2025-05-26 14:41:50,007] [INFO] [config.py:1007:print]   pipeline ..................... {'stages': 'auto', 'partition': 'best', 'seed_layers': False, 'activation_checkpoint_interval': 0, 'pipe_partitioned': True, 'grad_partitioned': True}
[2025-05-26 14:41:50,007] [INFO] [config.py:1007:print]   pld_enabled .................. False
[2025-05-26 14:41:50,007] [INFO] [config.py:1007:print]   pld_params ................... False
[2025-05-26 14:41:50,007] [INFO] [config.py:1007:print]   prescale_gradients ........... False
[2025-05-26 14:41:50,007] [INFO] [config.py:1007:print]   scheduler_name ............... None
[2025-05-26 14:41:50,007] [INFO] [config.py:1007:print]   scheduler_params ............. None
[2025-05-26 14:41:50,007] [INFO] [config.py:1007:print]   seq_parallel_communication_data_type  torch.float32
[2025-05-26 14:41:50,007] [INFO] [config.py:1007:print]   sparse_attention ............. None
[2025-05-26 14:41:50,007] [INFO] [config.py:1007:print]   sparse_gradients_enabled ..... False
[2025-05-26 14:41:50,007] [INFO] [config.py:1007:print]   steps_per_print .............. 10
[2025-05-26 14:41:50,007] [INFO] [config.py:1007:print]   tensor_parallel_config ....... dtype=torch.float16 autotp_size=0 tp_overlap_comm=False tensor_parallel=TPConfig(tp_size=1, tp_grain_size=1, mpu=None, tp_group=None) injection_policy_tuple=None keep_module_on_host=False replace_with_kernel_inject=False
[2025-05-26 14:41:50,007] [INFO] [config.py:1007:print]   timers_config ................ enabled=True synchronized=True
[2025-05-26 14:41:50,007] [INFO] [config.py:1007:print]   train_batch_size ............. 64
[2025-05-26 14:41:50,007] [INFO] [config.py:1007:print]   train_micro_batch_size_per_gpu  1
[2025-05-26 14:41:50,007] [INFO] [config.py:1007:print]   use_data_before_expert_parallel_  False
[2025-05-26 14:41:50,007] [INFO] [config.py:1007:print]   use_node_local_storage ....... False
[2025-05-26 14:41:50,007] [INFO] [config.py:1007:print]   wall_clock_breakdown ......... False
[2025-05-26 14:41:50,007] [INFO] [config.py:1007:print]   weight_quantization_config ... None
[2025-05-26 14:41:50,007] [INFO] [config.py:1007:print]   world_size ................... 8
[2025-05-26 14:41:50,007] [INFO] [config.py:1007:print]   zero_allow_untested_optimizer  False
[2025-05-26 14:41:50,007] [INFO] [config.py:1007:print]   zero_config .................. stage=3 contiguous_gradients=True reduce_scatter=True reduce_bucket_size=500000000 use_multi_rank_bucket_allreduce=True allgather_partitions=True allgather_bucket_size=500000000 overlap_comm=True load_from_fp32_weights=True elastic_checkpoint=False offload_param=DeepSpeedZeroOffloadParamConfig(device='none', nvme_path=None, buffer_count=5, buffer_size=100000000, max_in_cpu=1000000000, pin_memory=False) offload_optimizer=DeepSpeedZeroOffloadOptimizerConfig(device='none', nvme_path=None, buffer_count=4, pin_memory=False, pipeline_read=False, pipeline_write=False, fast_init=False, ratio=1.0) sub_group_size=1000000000 cpu_offload_param=None cpu_offload_use_pin_memory=None cpu_offload=None prefetch_bucket_size=30000000 param_persistence_threshold=10000 model_persistence_threshold=9223372036854775807 max_live_parameters=30000000 max_reuse_distance=1000000000 gather_16bit_weights_on_model_save=True module_granularity_threshold=0 use_all_reduce_for_fetch_params=False stage3_gather_fp16_weights_on_model_save=False ignore_unused_parameters=True legacy_stage1=False round_robin_gradients=False zero_hpz_partition_size=1 zero_quantized_weights=False zero_quantized_nontrainable_weights=False zero_quantized_gradients=False zeropp_loco_param=None mics_shard_size=-1 mics_hierarchical_params_gather=False memory_efficient_linear=False pipeline_loading_checkpoint=False override_module_apply=True log_trace_cache_warnings=False
[2025-05-26 14:41:50,007] [INFO] [config.py:1007:print]   zero_enabled ................. True
[2025-05-26 14:41:50,007] [INFO] [config.py:1007:print]   zero_force_ds_cpu_optimizer .. True
[2025-05-26 14:41:50,007] [INFO] [config.py:1007:print]   zero_optimization_stage ...... 3
[2025-05-26 14:41:50,007] [INFO] [config.py:993:print_user_config]   json = {
    "train_batch_size": 64, 
    "train_micro_batch_size_per_gpu": 1, 
    "gradient_accumulation_steps": 8, 
    "steps_per_print": 10, 
    "zero_optimization": {
        "stage": 3, 
        "offload_param": {
            "device": "none"
        }, 
        "offload_optimizer": {
            "device": "none"
        }, 
        "param_persistence_threshold": 1.000000e+04, 
        "max_live_parameters": 3.000000e+07, 
        "prefetch_bucket_size": 3.000000e+07, 
        "memory_efficient_linear": false, 
        "gather_16bit_weights_on_model_save": true
    }, 
    "gradient_clipping": 1.0, 
    "prescale_gradients": false, 
    "wall_clock_breakdown": false, 
    "hybrid_engine": {
        "enabled": false, 
        "max_out_tokens": 512, 
        "inference_tp_size": 1, 
        "release_inference_cache": false, 
        "pin_parameters": true, 
        "tp_gather_partition_size": 8
    }, 
    "bf16": {
        "enabled": true
    }
}
***** Running training *****
[2025-05-26 14:43:12,002] [INFO] [logging.py:107:log_dist] [Rank 0] step=10, skipped=0, lr=[1e-05, 1e-05], mom=[(0.9, 0.95), (0.9, 0.95)]
[2025-05-26 14:43:12,002] [INFO] [timer.py:264:stop] epoch=0/micro_step=80/global_step=10, RunningAvgSamplesPerSec=9.556703229044809, CurrSamplesPerSec=10.39685946239257, MemAllocated=1.86GB, MaxMemAllocated=3.46GB
[2025-05-26 14:44:24,613] [INFO] [logging.py:107:log_dist] [Rank 0] step=20, skipped=0, lr=[1e-05, 1e-05], mom=[(0.9, 0.95), (0.9, 0.95)]
[2025-05-26 14:44:24,613] [INFO] [timer.py:264:stop] epoch=0/micro_step=160/global_step=20, RunningAvgSamplesPerSec=9.37938227439661, CurrSamplesPerSec=8.205715626714705, MemAllocated=1.86GB, MaxMemAllocated=3.46GB
[2025-05-26 14:45:04,810] [INFO] [logging.py:107:log_dist] [Rank 0] step=30, skipped=0, lr=[1e-05, 1e-05], mom=[(0.9, 0.95), (0.9, 0.95)]
[2025-05-26 14:45:04,811] [INFO] [timer.py:264:stop] epoch=0/micro_step=240/global_step=30, RunningAvgSamplesPerSec=11.276623221304217, CurrSamplesPerSec=13.343929523095127, MemAllocated=1.86GB, MaxMemAllocated=3.48GB
[2025-05-26 14:46:11,384] [INFO] [logging.py:107:log_dist] [Rank 0] step=40, skipped=0, lr=[1e-05, 1e-05], mom=[(0.9, 0.95), (0.9, 0.95)]
[2025-05-26 14:46:11,384] [INFO] [timer.py:264:stop] epoch=0/micro_step=320/global_step=40, RunningAvgSamplesPerSec=10.993007255570975, CurrSamplesPerSec=9.751216410935061, MemAllocated=1.86GB, MaxMemAllocated=3.48GB
[2025-05-26 14:47:22,442] [INFO] [logging.py:107:log_dist] [Rank 0] step=50, skipped=0, lr=[1e-05, 1e-05], mom=[(0.9, 0.95), (0.9, 0.95)]
[2025-05-26 14:47:22,442] [INFO] [timer.py:264:stop] epoch=0/micro_step=400/global_step=50, RunningAvgSamplesPerSec=10.627451474074665, CurrSamplesPerSec=9.574698218891845, MemAllocated=1.86GB, MaxMemAllocated=3.48GB
[2025-05-26 14:48:29,734] [INFO] [logging.py:107:log_dist] [Rank 0] step=60, skipped=0, lr=[1e-05, 1e-05], mom=[(0.9, 0.95), (0.9, 0.95)]
[2025-05-26 14:48:29,734] [INFO] [timer.py:264:stop] epoch=0/micro_step=480/global_step=60, RunningAvgSamplesPerSec=10.517911878118918, CurrSamplesPerSec=9.931604335344115, MemAllocated=1.86GB, MaxMemAllocated=3.48GB
[2025-05-26 14:49:40,236] [INFO] [logging.py:107:log_dist] [Rank 0] step=70, skipped=0, lr=[1e-05, 1e-05], mom=[(0.9, 0.95), (0.9, 0.95)]
[2025-05-26 14:49:40,237] [INFO] [timer.py:264:stop] epoch=0/micro_step=560/global_step=70, RunningAvgSamplesPerSec=10.36077776757022, CurrSamplesPerSec=10.803676672920048, MemAllocated=1.86GB, MaxMemAllocated=3.48GB
Saving model to "/aifs4su/hansirui_1st/jiayi/setting3-imdb/Qwen1.5-0.5B/Qwen1.5-0.5B-s3-Q1-5000" ...
Saving 16-bit model...
[2025-05-26 14:50:39,125] [INFO] [torch_checkpoint_engine.py:33:commit] [Torch] Checkpoint global_step78 is ready now!
[2025-05-26 14:50:39,125] [INFO] [torch_checkpoint_engine.py:33:commit] [Torch] Checkpoint global_step78 is ready now!
[2025-05-26 14:50:39,125] [INFO] [torch_checkpoint_engine.py:33:commit] [Torch] Checkpoint global_step78 is ready now!
[2025-05-26 14:50:39,125] [INFO] [torch_checkpoint_engine.py:33:commit] [Torch] Checkpoint global_step78 is ready now!
[2025-05-26 14:50:39,125] [INFO] [torch_checkpoint_engine.py:33:commit] [Torch] Checkpoint global_step78 is ready now!
[2025-05-26 14:50:39,125] [INFO] [torch_checkpoint_engine.py:33:commit] [Torch] Checkpoint global_step78 is ready now!
[2025-05-26 14:50:39,125] [INFO] [logging.py:107:log_dist] [Rank 0] [Torch] Checkpoint global_step78 is about to be saved!
[2025-05-26 14:50:39,125] [INFO] [torch_checkpoint_engine.py:33:commit] [Torch] Checkpoint global_step78 is ready now!
[2025-05-26 14:50:39,126] [INFO] [engine.py:3860:save_16bit_model] Saving model weights to /aifs4su/hansirui_1st/jiayi/setting3-imdb/Qwen1.5-0.5B/Qwen1.5-0.5B-s3-Q1-5000/pytorch_model.bin, tag: global_step78
[2025-05-26 14:50:39,126] [INFO] [torch_checkpoint_engine.py:21:save] [Torch] Saving /aifs4su/hansirui_1st/jiayi/setting3-imdb/Qwen1.5-0.5B/Qwen1.5-0.5B-s3-Q1-5000/pytorch_model.bin...
[2025-05-26 14:50:39,976] [INFO] [torch_checkpoint_engine.py:23:save] [Torch] Saved /aifs4su/hansirui_1st/jiayi/setting3-imdb/Qwen1.5-0.5B/Qwen1.5-0.5B-s3-Q1-5000/pytorch_model.bin.
[2025-05-26 14:50:39,976] [INFO] [torch_checkpoint_engine.py:33:commit] [Torch] Checkpoint global_step78 is ready now!
Model saved!
[2025-05-26 14:50:41,378] [INFO] [launch.py:351:main] Process 3396238 exits successfully.
[2025-05-26 14:50:41,378] [INFO] [launch.py:351:main] Process 3396235 exits successfully.
wandb: 
wandb: 🚀 View run imdb-Qwen1.5-0.5B-s3-Q1-5000 at: https://wandb.ai/xtom/Inverse_Alignment_IMDb/runs/cf91tfa2
wandb: Find logs at: ../../../../../../aifs4su/hansirui_1st/jiayi/setting3-imdb/Qwen1.5-0.5B/Qwen1.5-0.5B-s3-Q1-5000/wandb/run-20250526_144150-cf91tfa2/logs
[2025-05-26 14:50:42,379] [INFO] [launch.py:351:main] Process 3396236 exits successfully.
[2025-05-26 14:50:42,379] [INFO] [launch.py:351:main] Process 3396232 exits successfully.
[2025-05-26 14:50:42,379] [INFO] [launch.py:351:main] Process 3396234 exits successfully.
[2025-05-26 14:50:42,379] [INFO] [launch.py:351:main] Process 3396237 exits successfully.
[2025-05-26 14:50:42,380] [INFO] [launch.py:351:main] Process 3396233 exits successfully.
[2025-05-26 14:50:43,380] [INFO] [launch.py:351:main] Process 3396231 exits successfully.