Skip to main content
eleutherai
Projects
neox
Run groups
fdaTaxD5KxESYpsAhuuY5V_1haf437g
Logs
Group
Workspace
Runs
Logs
Files
Artifacts

Chilli's group workspace

Timestamps visible
2023-08-03 10:56:51
Using ./extensions/ as PyTorch extensions root...
2023-08-03 10:56:51
No modifications detected for re-loaded extension module utils, skipping build step...
2023-08-03 10:56:51
Loading extension module utils...
2023-08-03 10:56:55
[2023-08-03 10:56:53,336] [INFO] [engine.py:1551:_load_checkpoint] rank: 24 loading checkpoint: /fsx/lintangsutawika/checkpoints/temp_neox_models/global_step38000/mp_rank_00_model_states.pt
2023-08-03 10:58:51
successfully loaded 64 ZeRO state_dicts for rank 24
2023-08-03 10:59:20
loading 64 zero partition checkpoints for rank 24
2023-08-03 10:59:40
WARNING: shuffle index length (162165685) is not equal to sample index length (162165686)
2023-08-03 10:59:42
WARNING: shuffle index length (162165685) is not equal to sample index length (162165686)
2023-08-03 10:59:44
WARNING: shuffle index length (162165685) is not equal to sample index length (162165686)
2023-08-03 10:59:46
> RANK 24 elapsed time for building blendable dataset indices: 0.59 (sec)
2023-08-03 10:59:46
> RANK 24 elapsed time for building blendable dataset indices: 1.00 (sec)
2023-08-03 10:59:48
> RANK 24 elapsed time for building blendable dataset indices: 1.07 (sec)
2023-08-03 11:00:21
[2023-08-03 11:00:19,856] [INFO] [engine.py:1805:_copy_recovery_script] creating recovery script /fsx/lintangsutawika/checkpoints/temp_neox_models/zero_to_fp32.py
2023-08-03 11:00:21
[2023-08-03 11:00:19,869] [INFO] [engine.py:1818:_save_zero_checkpoint] zero checkpoint saved /fsx/lintangsutawika/checkpoints/temp_neox_models/global_step38001/zero_pp_rank_24_mp_rank_00_optim_states.pt
2023-08-03 11:00:35
[2023-08-03 11:00:35,180] [INFO] [engine.py:1805:_copy_recovery_script] creating recovery script /fsx/lintangsutawika/checkpoints/temp_neox_models/zero_to_fp32.py
2023-08-03 11:00:35
[2023-08-03 11:00:35,188] [INFO] [engine.py:1818:_save_zero_checkpoint] zero checkpoint saved /fsx/lintangsutawika/checkpoints/temp_neox_models/global_step38002/zero_pp_rank_24_mp_rank_00_optim_states.pt