megatron_utils.py in load_checkpoint: args.iteration = state_dict['iteration'] TypeError: 'NoneType' object is not subscriptable
Open
- Dominant language
- Python
- Stars
- 450
- Forks
- 40
- PR merge metrics
- No merged PRs in 30d
Description
Has anyone encountered this issue? I got this error when running train_rlhf_llama.sh. here is my .yaml and .log
![Uploading QQ_1735835251973.png…]()
Contributor guide
No contributing guide indexed for this repository
Research direction
Start with megatron_utils.py in load_checkpoint and inspect where state_dict is obtained before state_dict['iteration'] is accessed. Use the referenced train_rlhf_llama.sh run and review the attached YAML and log; done means identifying why the checkpoint state is None and confirming the run no longer raises this TypeError.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- python
- Domain
- machine-learning
- Issue type
- Bug
- Difficulty
- 3/5
- Estimated time
- 1-2 days
- Activity status
- Stale
- Clarity
- Needs clarification
- Newbie friendliness
- 20/100