Skip to main content

Eleutherai-oslo's group workspace

Timestamps visible
2023-04-20 01:21:30
[2023-04-20 01:21:28,730] [INFO] [distributed.py:46:init_distributed] Initializing torch distributed with backend: nccl
2023-04-20 01:21:30
FAILED !!!
2023-04-20 01:21:32
Traceback (most recent call last):
2023-04-20 01:21:32
  File "/fsx/polyglot.train/gpt-neox/train.py", line 27, in <module>
2023-04-20 01:21:32
    pretrain(neox_args=neox_args)
2023-04-20 01:21:32
  File "/fsx/polyglot.train/gpt-neox/megatron/training.py", line 104, in pretrain
2023-04-20 01:21:32
    model, optimizer, lr_scheduler = setup_model_and_optimizer(
2023-04-20 01:21:32
  File "/fsx/polyglot.train/gpt-neox/megatron/training.py", line 440, in setup_model_and_optimizer
2023-04-20 01:21:32
    optimizer, param_groups = get_optimizer(model=model, neox_args=neox_args)
2023-04-20 01:21:32
UnboundLocalError: local variable 'model' referenced before assignment