Can't this model be trained using multi nodes and multiple cards ?

Aperta

#3 aperta il 4/5/2023 2 commenti 0 reazioni 0 assegnatari Vedi su GitHub

Nessuno ha ancora preso questa issue.

Valutazione

Difficoltà: 5/5
Tempo stimato: Più di una settimana
Idoneità per principianti: 25/100

Tipo di issue: Funzionalità
Chiarezza: Da chiarire
Stato di attività: Ferma
Stack tecnologico: python
Ambito: distributed-systems, machine-learning

Direzione di ricerca

Start with train.py and the multinode_runner.py and DeepSpeed launcher commands shown in the logs. Reproduce the 16B-model run across the listed nodes and inspect where memory use remains per GPU. Done would require a documented or implemented way to train the model across multiple machines and cards without OOM.

Scritto dal modello di indicizzazione a partire dal testo della issue.

Descrizione

My machine has a single node 4-card 16G graphics memory, and running the 16B model with multiple nodes will result in OOM regardless of how the number of nodes is set. Can this model be trained with multiple machines and cards
Here are some logs of the training：

2023-05-04 16:21:02,954] [INFO] [multinode_runner.py:65:get_cmd] Running on the following workers: c05r2n11,c05r2n16,c05r2n19,c05r3n02,c06r4n06,c06r4n09,c06r4n12,c06r4n17
[2023-05-04 16:21:02,955] [INFO] [runner.py:453:main] cmd = pdsh -f 1024 -w c05r2n11,c05r2n16,c05r2n19,c05r3n02,c06r4n06,c06r4n09,c06r4n12,c06r4n17 export NCCL_SOCKET_IFNAME=ib0; export UCX_MAX_EAGER_LANES=4; export UCX_MAX_RNDV_LANES=4; export UCX_ZCOPY_THRESH=auto; export UCX_RNDV_THRESH=auto; export UCX_DC_MLX5_NUM_DCI=16; export NCCL_IB_HCA=mlx5_0; export NCCL_DEBUG=info; export PYTHONPATH=/work/home/actvg1ue59/CodeGen-main; cd /work/home/actvg1ue59/CodeGen-main; /work/home/actvg1ue59/miniconda3/envs/torch/bin/python -u -m deepspeed.launcher.launch --world_info=eyJjMDVyMm4xMSI6IFswLCAxLCAyLCAzXSwgImMwNXIybjE2IjogWzAsIDEsIDIsIDNdLCAiYzA1cjJuMTkiOiBbMCwgMSwgMiwgM10sICJjMDVyM24wMiI6IFswLCAxLCAyLCAzXSwgImMwNnI0bjA2IjogWzAsIDEsIDIsIDNdLCAiYzA2cjRuMDkiOiBbMCwgMSwgMiwgM10sICJjMDZyNG4xMiI6IFswLCAxLCAyLCAzXSwgImMwNnI0bjE3IjogWzAsIDEsIDIsIDNdfQ== --node_rank=%n --master_addr=c05r2n11 --master_port=9901 train.py
c05r3n02: Warning: Permanently added 'c05r3n02,10.3.5.43' (ECDSA) to the list of known hosts.^M
c05r2n16: Warning: Permanently added 'c05r2n16,10.3.5.37' (ECDSA) to the list of known hosts.^M
c05r2n19: Warning: Permanently added 'c05r2n19,10.3.5.40' (ECDSA) to the list of known hosts.^M
c06r4n06: Warning: Permanently added 'c06r4n06,10.3.6.67' (ECDSA) to the list of known hosts.^M
c06r4n09: Warning: Permanently added 'c06r4n09,10.3.6.70' (ECDSA) to the list of known hosts.^M
c06r4n12: Warning: Permanently added 'c06r4n12,10.3.6.73' (ECDSA) to the list of known hosts.^M
c06r4n17: Warning: Permanently added 'c06r4n17,10.3.6.78' (ECDSA) to the list of known hosts.^M
c05r2n11: Currently Loaded Modulefiles:
c05r2n11: 1) compiler/devtoolset/7.3.1 3) compiler/dtk/22.10.1
c05r2n11: 2) mpi/hpcx/gcc-7.3.1
c05r2n11: [2023-05-04 16:21:07,883] [INFO] [launch.py:96:main] 0 NCCL_SOCKET_IFNAME=ib0
c05r2n11: [2023-05-04 16:21:07,883] [INFO] [launch.py:96:main] 0 NCCL_IB_HCA=mlx5_0
c05r2n11: [2023-05-04 16:21:07,883] [INFO] [launch.py:96:main] 0 NCCL_DEBUG=info
c05r2n11: [2023-05-04 16:21:07,883] [INFO] [launch.py:103:main] WORLD INFO DICT: {'c05r2n11': [0, 1, 2, 3], 'c05r2n16': [0, 1, 2, 3], 'c05r2n19': [0, 1, 2, 3], 'c05r3n02': [0, 1, 2, 3], 'c06r4n06': [0, 1, 2, 3], 'c06r4n09': [0, 1, 2, 3], 'c06r4n12': [0, 1, 2, 3], 'c06r4n17': [0, 1, 2, 3]}
c05r2n11: [2023-05-04 16:21:07,883] [INFO] [launch.py:109:main] nnodes=8, num_local_procs=4, node_rank=0
c05r2n11: [2023-05-04 16:21:07,884] [INFO] [launch.py:122:main] global_rank_mapping=defaultdict(<class 'list'>, {'c05r2n11': [0, 1, 2, 3], 'c05r2n16': [4, 5, 6, 7], 'c05r2n19': [8, 9, 10, 11], 'c05r3n02': [12, 13, 14, 15], 'c06r4n06': [16, 17, 18, 19], 'c06r4n09': [20, 21, 22, 23], 'c06r4n12': [24, 25, 26, 27], 'c06r4n17': [28, 29, 30, 31]})
c05r2n11: [2023-05-04 16:21:07,884] [INFO] [launch.py:123:main] dist_world_size=32
c05r2n11: [2023-05-04 16:21:07,884] [INFO] [launch.py:125:main] Setting CUDA_VISIBLE_DEVICES=0,1,2,3
c05r2n16: [2023-05-04 16:21:36,542] [INFO] [launch.py:109:main] nnodes=8, num_local_procs=4, node_rank=1
c06r4n17: [2023-05-04 16:21:36,534] [INFO] [launch.py:96:main] 7 NCCL_IB_HCA=mlx5_0
c05r2n19: [2023-05-04 16:21:36,531] [INFO] [launch.py:96:main] 2 NCCL_IB_HCA=mlx5_0
c06r4n12: [2023-05-04 16:21:36,538] [INFO] [launch.py:96:main] 6 NCCL_DEBUG=info
c05r2n16: [2023-05-04 16:21:36,542] [INFO] [launch.py:122:main] global_rank_mapping=defaultdict(<class 'list'>, {'c05r2n11': [0, 1, 2, 3], 'c05r2n16': [4, 5, 6, 7], 'c05r2n19': [8, 9, 10, 11], 'c05r3n02': [12, 13, 14, 15], 'c06r4n06': [16, 17, 18, 19], 'c06r4n09': [20, 21, 22, 23], 'c06r4n12': [24, 25, 26, 27], 'c06r4n17': [28, 29, 30, 31]})
c06r4n17: [2023-05-04 16:21:36,534] [INFO] [launch.py:96:main] 7 NCCL_DEBUG=info
c06r4n12: [2023-05-04 16:21:36,538] [INFO] [launch.py:103:main] WORLD INFO DICT: {'c05r2n11': [0, 1, 2, 3], 'c05r2n16': [0, 1, 2, 3], 'c05r2n19': [0, 1, 2, 3], 'c05r3n02': [0, 1, 2, 3], 'c06r4n06': [0, 1, 2, 3], 'c06r4n09': [0, 1, 2, 3], 'c06r4n12': [0, 1, 2, 3], 'c06r4n17': [0, 1, 2, 3]}
c05r2n19: [2023-05-04 16:21:36,531] [INFO] [launch.py:96:main] 2 NCCL_DEBUG=info
c06r4n06: [2023-05-04 16:21:36,543] [INFO] [launch.py:109:main] nnodes=8, num_local_procs=4, node_rank=4
c06r4n09: [2023-05-04 16:21:36,546] [INFO] [launch.py:96:main] 5 NCCL_IB_HCA=mlx5_0
c05r3n02: [2023-05-04 16:21:36,531] [INFO] [launch.py:96:main] 3 NCCL_IB_HCA=mlx5_0
c06r4n09: [2023-05-04 16:21:36,546] [INFO] [launch.py:96:main] 5 NCCL_DEBUG=info
c05r2n19: [2023-05-04 16:21:36,531] [INFO] [launch.py:103:main] WORLD INFO DICT: {'c05r2n11': [0, 1, 2, 3], 'c05r2n16': [0, 1, 2, 3], 'c05r2n19': [0, 1, 2, 3], 'c05r3n02': [0, 1, 2, 3], 'c06r4n06': [0, 1, 2, 3], 'c06r4n09': [0, 1, 2, 3], 'c06r4n12': [0, 1, 2, 3], 'c06r4n17': [0, 1, 2, 3]}
c06r4n06: [2023-05-04 16:21:36,543] [INFO] [launch.py:122:main] global_rank_mapping=defaultdict(<class 'list'>, {'c05r2n11': [0, 1, 2, 3], 'c05r2n16': [4, 5, 6, 7], 'c05r2n19': [8, 9, 10, 11], 'c05r3n02': [12, 13, 14, 15], 'c06r4n06': [16, 17, 18, 19], 'c06r4n09': [20, 21, 22, 23], 'c06r4n12': [24, 25, 26, 27], 'c06r4n17': [28, 29, 30, 31]})
c06r4n12: [2023-05-04 16:21:36,538] [INFO] [launch.py:109:main] nnodes=8, num_local_procs=4, node_rank=6
c06r4n12: [2023-05-04 16:21:36,538] [INFO] [launch.py:122:main] global_rank_mapping=defaultdict(<class 'list'>, {'c05r2n11': [0, 1, 2, 3], 'c05r2n16': [4, 5, 6, 7], 'c05r2n19': [8, 9, 10, 11], 'c05r3n02': [12, 13, 14, 15], 'c06r4n06': [16, 17, 18, 19], 'c06r4n09': [20, 21, 22, 23], 'c06r4n12': [24, 25, 26, 27], 'c06r4n17': [28, 29, 30, 31]})
c05r2n16: [2023-05-04 16:21:36,542] [INFO] [launch.py:123:main] dist_world_size=32
c05r2n16: [2023-05-04 16:21:36,542] [INFO] [launch.py:125:main] Setting CUDA_VISIBLE_DEVICES=0,1,2,3
c06r4n06: [2023-05-04 16:21:36,543] [INFO] [launch.py:123:main] dist_world_size=32
c06r4n12: [2023-05-04 16:21:36,538] [INFO] [launch.py:123:main] dist_world_size=32
c06r4n06: [2023-05-04 16:21:36,543] [INFO] [launch.py:125:main] Setting CUDA_VISIBLE_DEVICES=0,1,2,3
c06r4n12: [2023-05-04 16:21:36,538] [INFO] [launch.py:125:main] Setting CUDA_VISIBLE_DEVICES=0,1,2,3
c05r3n02: [2023-05-04 16:21:36,531] [INFO] [launch.py:96:main] 3 NCCL_DEBUG=info
c05r3n02: [2023-05-04 16:21:36,531] [INFO] [launch.py:103:main] WORLD INFO DICT: {'c05r2n11': [0, 1, 2, 3], 'c05r2n16': [0, 1, 2, 3], 'c05r2n19': [0, 1, 2, 3], 'c05r3n02': [0, 1, 2, 3], 'c06r4n06': [0, 1, 2, 3], 'c06r4n09': [0, 1, 2, 3], 'c06r4n12': [0, 1, 2, 3], 'c06r4n17': [0, 1, 2, 3]}
c05r2n19: [2023-05-04 16:21:36,531] [INFO] [launch.py:109:main] nnodes=8, num_local_procs=4, node_rank=2
c06r4n17: [2023-05-04 16:21:36,534] [INFO] [launch.py:103:main] WORLD INFO DICT: {'c05r2n11': [0, 1, 2, 3], 'c05r2n16': [0, 1, 2, 3], 'c05r2n19': [0, 1, 2, 3], 'c05r3n02': [0, 1, 2, 3], 'c06r4n06': [0, 1, 2, 3], 'c06r4n09': [0, 1, 2, 3], 'c06r4n12': [0, 1, 2, 3], 'c06r4n17': [0, 1, 2, 3]}
c05r2n19: [2023-05-04 16:21:36,531] [INFO] [launch.py:122:main] global_rank_mapping=defaultdict(<class 'list'>, {'c05r2n11': [0, 1, 2, 3], 'c05r2n16': [4, 5, 6, 7], 'c05r2n19': [8, 9, 10, 11], 'c05r3n02': [12, 13, 14, 15], 'c06r4n06': [16, 17, 18, 19], 'c06r4n09': [20, 21, 22, 23], 'c06r4n12': [24, 25, 26, 27], 'c06r4n17': [28, 29, 30, 31]})
c05r2n19: [2023-05-04 16:21:36,531] [INFO] [launch.py:123:main] dist_world_size=32
c05r2n19: [2023-05-04 16:21:36,531] [INFO] [launch.py:125:main] Setting CUDA_VISIBLE_DEVICES=0,1,2,3
c05r3n02: [2023-05-04 16:21:36,531] [INFO] [launch.py:109:main] nnodes=8, num_local_procs=4, node_rank=3
c06r4n17: [2023-05-04 16:21:36,534] [INFO] [launch.py:109:main] nnodes=8, num_local_procs=4, node_rank=7
c05r3n02: [2023-05-04 16:21:36,531] [INFO] [launch.py:122:main] global_rank_mapping=defaultdict(<class 'list'>, {'c05r2n11': [0, 1, 2, 3], 'c05r2n16': [4, 5, 6, 7], 'c05r2n19': [8, 9, 10, 11], 'c05r3n02': [12, 13, 14, 15], 'c06r4n06': [16, 17, 18, 19], 'c06r4n09': [20, 21, 22, 23], 'c06r4n12': [24, 25, 26, 27], 'c06r4n17': [28, 29, 30, 31]})
c06r4n17: [2023-05-04 16:21:36,534] [INFO] [launch.py:122:main] global_rank_mapping=defaultdict(<class 'list'>, {'c05r2n11': [0, 1, 2, 3], 'c05r2n16': [4, 5, 6, 7], 'c05r2n19': [8, 9, 10, 11], 'c05r3n02': [12, 13, 14, 15], 'c06r4n06': [16, 17, 18, 19], 'c06r4n09': [20, 21, 22, 23], 'c06r4n12': [24, 25, 26, 27], 'c06r4n17': [28, 29, 30, 31]})
c06r4n17: [2023-05-04 16:21:36,534] [INFO] [launch.py:123:main] dist_world_size=32
c05r3n02: [2023-05-04 16:21:36,532] [INFO] [launch.py:123:main] dist_world_size=32
c06r4n17: [2023-05-04 16:21:36,534] [INFO] [launch.py:125:main] Setting CUDA_VISIBLE_DEVICES=0,1,2,3
c05r3n02: [2023-05-04 16:21:36,532] [INFO] [launch.py:125:main] Setting CUDA_VISIBLE_DEVICES=0,1,2,3
c06r4n09: [2023-05-04 16:21:36,555] [INFO] [launch.py:103:main] WORLD INFO DICT: {'c05r2n11': [0, 1, 2, 3], 'c05r2n16': [0, 1, 2, 3], 'c05r2n19': [0, 1, 2, 3], 'c05r3n02': [0, 1, 2, 3], 'c06r4n06': [0, 1, 2, 3], 'c06r4n09': [0, 1, 2, 3], 'c06r4n12': [0, 1, 2, 3], 'c06r4n17': [0, 1, 2, 3]}
c06r4n09: [2023-05-04 16:21:36,555] [INFO] [launch.py:109:main] nnodes=8, num_local_procs=4, node_rank=5
c05r2n16: [2023-05-04 16:21:36,542] [INFO] [launch.py:109:main] nnodes=8, num_local_procs=4, node_rank=1
c06r4n17: [2023-05-04 16:21:36,534] [INFO] [launch.py:96:main] 7 NCCL_IB_HCA=mlx5_0
c05r2n19: [2023-05-04 16:21:36,531] [INFO] [launch.py:96:main] 2 NCCL_IB_HCA=mlx5_0
c06r4n12: [2023-05-04 16:21:36,538] [INFO] [launch.py:96:main] 6 NCCL_DEBUG=info
c05r2n16: [2023-05-04 16:21:36,542] [INFO] [launch.py:122:main] global_rank_mapping=defaultdict(<class 'list'>, {'c05r2n11': [0, 1, 2, 3], 'c05r2n16': [4, 5, 6, 7], 'c05r2n19': [8, 9, 10, 11], 'c05r3n02': [12, 13, 14, 15], 'c06r4n06': [16, 17, 18, 19], 'c06r4n09': [20, 21, 22, 23], 'c06r4n12': [24, 25, 26, 27], 'c06r4n17': [28, 29, 30, 31]})
c06r4n17: [2023-05-04 16:21:36,534] [INFO] [launch.py:96:main] 7 NCCL_DEBUG=info
c06r4n12: [2023-05-04 16:21:36,538] [INFO] [launch.py:103:main] WORLD INFO DICT: {'c05r2n11': [0, 1, 2, 3], 'c05r2n16': [0, 1, 2, 3], 'c05r2n19': [0, 1, 2, 3], 'c05r3n02': [0, 1, 2, 3], 'c06r4n06': [0, 1, 2, 3], 'c06r4n09': [0, 1, 2, 3], 'c06r4n12': [0, 1, 2, 3], 'c06r4n17': [0, 1, 2, 3]}
c05r2n19: [2023-05-04 16:21:36,531] [INFO] [launch.py:96:main] 2 NCCL_DEBUG=info
c06r4n06: [2023-05-04 16:21:36,543] [INFO] [launch.py:109:main] nnodes=8, num_local_procs=4, node_rank=4
c06r4n09: [2023-05-04 16:21:36,546] [INFO] [launch.py:96:main] 5 NCCL_IB_HCA=mlx5_0
c05r3n02: [2023-05-04 16:21:36,531] [INFO] [launch.py:96:main] 3 NCCL_IB_HCA=mlx5_0
c06r4n09: [2023-05-04 16:21:36,546] [INFO] [launch.py:96:main] 5 NCCL_DEBUG=info
c05r2n19: [2023-05-04 16:21:36,531] [INFO] [launch.py:103:main] WORLD INFO DICT: {'c05r2n11': [0, 1, 2, 3], 'c05r2n16': [0, 1, 2, 3], 'c05r2n19': [0, 1, 2, 3], 'c05r3n02': [0, 1, 2, 3], 'c06r4n06': [0, 1, 2, 3], 'c06r4n09': [0, 1, 2, 3], 'c06r4n12': [0, 1, 2, 3], 'c06r4n17': [0, 1, 2, 3]}
c06r4n06: [2023-05-04 16:21:36,543] [INFO] [launch.py:122:main] global_rank_mapping=defaultdict(<class 'list'>, {'c05r2n11': [0, 1, 2, 3], 'c05r2n16': [4, 5, 6, 7], 'c05r2n19': [8, 9, 10, 11], 'c05r3n02': [12, 13, 14, 15], 'c06r4n06': [16, 17, 18, 19], 'c06r4n09': [20, 21, 22, 23], 'c06r4n12': [24, 25, 26, 27], 'c06r4n17': [28, 29, 30, 31]})
c06r4n12: [2023-05-04 16:21:36,538] [INFO] [launch.py:109:main] nnodes=8, num_local_procs=4, node_rank=6
c06r4n12: [2023-05-04 16:21:36,538] [INFO] [launch.py:122:main] global_rank_mapping=defaultdict(<class 'list'>, {'c05r2n11': [0, 1, 2, 3], 'c05r2n16': [4, 5, 6, 7], 'c05r2n19': [8, 9, 10, 11], 'c05r3n02': [12, 13, 14, 15], 'c06r4n06': [16, 17, 18, 19], 'c06r4n09': [20, 21, 22, 23], 'c06r4n12': [24, 25, 26, 27], 'c06r4n17': [28, 29, 30, 31]})
c05r2n16: [2023-05-04 16:21:36,542] [INFO] [launch.py:123:main] dist_world_size=32
c05r2n16: [2023-05-04 16:21:36,542] [INFO] [launch.py:125:main] Setting CUDA_VISIBLE_DEVICES=0,1,2,3
c06r4n06: [2023-05-04 16:21:36,543] [INFO] [launch.py:123:main] dist_world_size=32
c06r4n12: [2023-05-04 16:21:36,538] [INFO] [launch.py:123:main] dist_world_size=32
c06r4n06: [2023-05-04 16:21:36,543] [INFO] [launch.py:125:main] Setting CUDA_VISIBLE_DEVICES=0,1,2,3
c06r4n12: [2023-05-04 16:21:36,538] [INFO] [launch.py:125:main] Setting CUDA_VISIBLE_DEVICES=0,1,2,3
c05r3n02: [2023-05-04 16:21:36,531] [INFO] [launch.py:96:main] 3 NCCL_DEBUG=info
c05r3n02: [2023-05-04 16:21:36,531] [INFO] [launch.py:103:main] WORLD INFO DICT: {'c05r2n11': [0, 1, 2, 3], 'c05r2n16': [0, 1, 2, 3], 'c05r2n19': [0, 1, 2, 3], 'c05r3n02': [0, 1, 2, 3], 'c06r4n06': [0, 1, 2, 3], 'c06r4n09': [0, 1, 2, 3], 'c06r4n12': [0, 1, 2, 3], 'c06r4n17': [0, 1, 2, 3]}
c05r2n19: [2023-05-04 16:21:36,531] [INFO] [launch.py:109:main] nnodes=8, num_local_procs=4, node_rank=2
c06r4n17: [2023-05-04 16:21:36,534] [INFO] [launch.py:103:main] WORLD INFO DICT: {'c05r2n11': [0, 1, 2, 3], 'c05r2n16': [0, 1, 2, 3], 'c05r2n19': [0, 1, 2, 3], 'c05r3n02': [0, 1, 2, 3], 'c06r4n06': [0, 1, 2, 3], 'c06r4n09': [0, 1, 2, 3], 'c06r4n12': [0, 1, 2, 3], 'c06r4n17': [0, 1, 2, 3]}
c05r2n19: [2023-05-04 16:21:36,531] [INFO] [launch.py:122:main] global_rank_mapping=defaultdict(<class 'list'>, {'c05r2n11': [0, 1, 2, 3], 'c05r2n16': [4, 5, 6, 7], 'c05r2n19': [8, 9, 10, 11], 'c05r3n02': [12, 13, 14, 15], 'c06r4n06': [16, 17, 18, 19], 'c06r4n09': [20, 21, 22, 23], 'c06r4n12': [24, 25, 26, 27], 'c06r4n17': [28, 29, 30, 31]})
c05r2n19: [2023-05-04 16:21:36,531] [INFO] [launch.py:123:main] dist_world_size=32
c05r2n19: [2023-05-04 16:21:36,531] [INFO] [launch.py:125:main] Setting CUDA_VISIBLE_DEVICES=0,1,2,3
c05r3n02: [2023-05-04 16:21:36,531] [INFO] [launch.py:109:main] nnodes=8, num_local_procs=4, node_rank=3
c06r4n17: [2023-05-04 16:21:36,534] [INFO] [launch.py:109:main] nnodes=8, num_local_procs=4, node_rank=7
c05r3n02: [2023-05-04 16:21:36,531] [INFO] [launch.py:122:main] global_rank_mapping=defaultdict(<class 'list'>, {'c05r2n11': [0, 1, 2, 3], 'c05r2n16': [4, 5, 6, 7], 'c05r2n19': [8, 9, 10, 11], 'c05r3n02': [12, 13, 14, 15], 'c06r4n06': [16, 17, 18, 19], 'c06r4n09': [20, 21, 22, 23], 'c06r4n12': [24, 25, 26, 27], 'c06r4n17': [28, 29, 30, 31]})
c06r4n17: [2023-05-04 16:21:36,534] [INFO] [launch.py:122:main] global_rank_mapping=defaultdict(<class 'list'>, {'c05r2n11': [0, 1, 2, 3], 'c05r2n16': [4, 5, 6, 7], 'c05r2n19': [8, 9, 10, 11], 'c05r3n02': [12, 13, 14, 15], 'c06r4n06': [16, 17, 18, 19], 'c06r4n09': [20, 21, 22, 23], 'c06r4n12': [24, 25, 26, 27], 'c06r4n17': [28, 29, 30, 31]})
c06r4n17: [2023-05-04 16:21:36,534] [INFO] [launch.py:123:main] dist_world_size=32
c05r3n02: [2023-05-04 16:21:36,532] [INFO] [launch.py:123:main] dist_world_size=32
c06r4n17: [2023-05-04 16:21:36,534] [INFO] [launch.py:125:main] Setting CUDA_VISIBLE_DEVICES=0,1,2,3
c05r3n02: [2023-05-04 16:21:36,532] [INFO] [launch.py:125:main] Setting CUDA_VISIBLE_DEVICES=0,1,2,3
c06r4n09: [2023-05-04 16:21:36,555] [INFO] [launch.py:103:main] WORLD INFO DICT: {'c05r2n11': [0, 1, 2, 3], 'c05r2n16': [0, 1, 2, 3], 'c05r2n19': [0, 1, 2, 3], 'c05r3n02': [0, 1, 2, 3], 'c06r4n06': [0, 1, 2, 3], 'c06r4n09': [0, 1, 2, 3], 'c06r4n12': [0, 1, 2, 3], 'c06r4n17': [0, 1, 2, 3]}
c06r4n09: [2023-05-04 16:21:36,555] [INFO] [launch.py:109:main] nnodes=8, num_local_procs=4, node_rank=5
c05r2n11: File "/work/home/actvg1ue59/miniconda3/envs/torch/lib/python3.9/site-packages/accelerate/utils/modeling.py", line 946, in load_checkpoint_in_model
c05r2n11: load_checkpoint_in_model(
c05r2n11: File "/work/home/actvg1ue59/miniconda3/envs/torch/lib/python3.9/site-packages/accelerate/utils/modeling.py", line 946, in load_checkpoint_in_model
c05r2n11: load_checkpoint_in_model(
c05r2n11: File "/work/home/actvg1ue59/miniconda3/envs/torch/lib/python3.9/site-packages/accelerate/utils/modeling.py", line 946, in load_checkpoint_in_model
c05r2n11: set_module_tensor_to_device(model, param_name, param_device, value=param, dtype=dtype)
c05r2n11: File "/work/home/actvg1ue59/miniconda3/envs/torch/lib/python3.9/site-packages/accelerate/utils/modeling.py", line 149, in set_module_tensor_to_device
c05r2n11: set_module_tensor_to_device(model, param_name, param_device, value=param, dtype=dtype)
c05r2n11: File "/work/home/actvg1ue59/miniconda3/envs/torch/lib/python3.9/site-packages/accelerate/utils/modeling.py", line 149, in set_module_tensor_to_device
c05r2n11: set_module_tensor_to_device(model, param_name, param_device, value=param, dtype=dtype)set_module_tensor_to_device(model, param_name, param_device, value=param, dtype=dtype)
c05r2n11:
c05r2n11: File "/work/home/actvg1ue59/miniconda3/envs/torch/lib/python3.9/site-packages/accelerate/utils/modeling.py", line 149, in set_module_tensor_to_device
c05r2n11: File "/work/home/actvg1ue59/miniconda3/envs/torch/lib/python3.9/site-packages/accelerate/utils/modeling.py", line 149, in set_module_tensor_to_device
c05r2n11: new_value = value.to(device)new_value = value.to(device)
c05r2n11:
c05r2n11: new_value = value.to(device)
c05r2n11: new_value = value.to(device)
c05r2n11: RuntimeErrorRuntimeError: : HIP out of memory. Tried to allocate 576.00 MiB (GPU 3; 15.98 GiB total capacity; 15.83 GiB already allocated; 0 bytes free; 15.84 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_HIP_ALLOC_CONFHIP out of memory. Tried to allocate 576.00 MiB (GPU 1; 15.98 GiB total capacity; 15.83 GiB already allocated; 0 bytes free; 15.84 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_HIP_ALLOC_CONF
c05r2n11: RuntimeError
c05r2n11: RuntimeError: : HIP out of memory. Tried to allocate 576.00 MiB (GPU 2; 15.98 GiB total capacity; 15.83 GiB already allocated; 0 bytes free; 15.84 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_HIP_ALLOC_CONFHIP out of memory. Tried to allocate 576.00 MiB (GPU 0; 15.98 GiB total capacity; 15.83 GiB already allocated; 0 bytes free; 15.84 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_HIP_ALLOC_CONF

Lingua principale: Python
Stelle: 268
Fork: 11
Metriche di merge delle PR: Nessuna PR unita negli ultimi 30g

Guida per i contributori

Nessuna guida per i contributori indicizzata per questo repository

Come iniziare

Leggi tutta la issue e poi la guida ai contributi del progetto.
Commenta sulla issue per dire che te ne occupi tu — evita che due persone facciano lo stesso lavoro.
Fai un fork del repository e lavora su un branch.
Apri una pull request che faccia riferimento al numero della issue.

Altre issue di salesforce/CodeGen2

Question of unk, bos, and eos tokens. Aperta

Difficoltà 4/5 3-5 giorni Idoneità per principianti 25/100

salesforce/CodeGen2#11 · Aperta il 20/7/2023
Question of eos token in codegen2-7B model Aperta

Difficoltà 3/5 1-2 giorni Idoneità per principianti 30/100

salesforce/CodeGen2#10 · Aperta il 27/6/2023
Will the training implementation of CodeGen 2 be released? Aperta

Difficoltà 5/5 Più di una settimana Idoneità per principianti 15/100

salesforce/CodeGen2#9 · Aperta il 15/6/2023
Is CodeGen2-16B the final and complete version? Aperta

Difficoltà 5/5 Più di una settimana Idoneità per principianti 15/100

salesforce/CodeGen2#7 · Aperta il 12/5/2023
How is it compared with the StarCoder Aperta

Difficoltà 5/5 Più di una settimana Idoneità per principianti 20/100

salesforce/CodeGen2#6 · 1 reazione · Aperta il 9/5/2023

Tutte le issue di salesforce/CodeGen2

Issue simili

docs: update workflows.md for slide deck Node test discovery Aperta

agent-ready documentation needs-triage

Difficoltà 1/5 1-3 ore Idoneità per principianti 88/100

microsoft/hve-core#3012 · Aperta il 25/9/2026
config_keys.rst says every result file PyBNF writes is alphabetical, but the credible-interval files and start_point.txt use declaration order Aperta

documentation

Difficoltà 1/5 Meno di un'ora Idoneità per principianti 91/100

lanl/PyBNF#932 · Aperta il 25/9/2026
workflow-status page template still says reusable workflows are "triggered only by workflow_call:" Aperta

Difficoltà 1/5 Meno di un'ora Idoneità per principianti 92/100

LearningCircuit/local-deep-research#6804 · Aperta il 24/9/2026
Add https://search.jeremyh.xyz/ Aperta

instance instance add

Difficoltà 1/5 Meno di un'ora Idoneità per principianti 72/100

searxng/searx-instances#939 · 1 commento · Aperta il 24/9/2026
Azure Sandboxes: Small and ExtraSmall tiers always fail to deploy (disk request exceeds tier maximum) Aperta

area-deployment area-integrations triage:bot-seen

Difficoltà 2/5 Mezza giornata Idoneità per principianti 86/100

microsoft/aspire#20459 · 1 commento · Aperta il 24/9/2026

Altre issue su Python