deepspeedai / deepspeedai/DeepSpeed
[BUG] RuntimeError: element 0 of tensors does not require grad and does not have a grad_fn
Open
@tohtana is already working on this.
Since Sep 6, 2023.
bug
training
- Dominant language
- Python
- Stars
- 43.1k
- Forks
- 5k
- Avg merge
- 4d 15h
- Merged PRs (30d)
- 112
Description
Hello，we meet an issue during using deepspeed pipeline engine. When we set the activation_checkpoint_interval=0 in the PipelineModule class, the structure works. But when we set it >0, the system will report
'RuntimeError: element 0 of tensors does not require grad and does not have a grad_fn'. I have checked the input_tensors of element 0, it looks like a hidden with (btach_size,seq_len,hidden_size)
How should I debug and solve the problem?
### Tasks
Contributor guide
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
Assessment
This issue has not been assessed yet.