Skip to content

Issues: NVIDIA/NeMo-Aligner

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

38 Open 7 Closed
38 Open 7 Closed
Author
Filter by author
Label
Filter by label
Use alt + click/return to exclude labels
or ⇧ + click/return for logical OR
Projects
Filter by project
Milestones
Filter by milestone
Assignee
Filter by who’s assigned
Sort
Sort by
Newest Oldest Most commented Least commented Recently updated Least recently updated Best match
Most reactions
👍 👎 😄 🎉 😕 ❤️ 🚀 👀

Issues list

Amend SPIN to be able to handle the cast of rollout_MBS < DP_size
#171 opened May 3, 2024 by trias702
Docker build failing. Also, is there a .nemo reward model file available? bug Something isn't working
#167 opened May 1, 2024 by rundiffusion
@JRD971000 @gshennvm
4
Can you please support context parallel?
#162 opened Apr 23, 2024 by DZ9
SFT val_check_interval should accept float input
#160 opened Apr 18, 2024 by AtsunoriFujita
SFT does not work max_steps bug Something isn't working
#159 opened Apr 18, 2024 by AtsunoriFujita
2
Can you support KTO?
#143 opened Apr 7, 2024 by lifan-yuan
1
cannot load reward model from SFT model because of missing keys bug Something isn't working
#137 opened Apr 1, 2024 by DZ9
8
SFT is broken with container 24.01.01 bug Something isn't working
#131 opened Mar 22, 2024 by odelalleau
1
SFT may crash if input data exceeds the context length bug Something isn't working
#127 opened Mar 15, 2024 by odelalleau
1
Support converting HF reward models to .nemo
#115 opened Feb 27, 2024 by odelalleau
1
random samplers keeps state bug Something isn't working
#107 opened Feb 14, 2024 by gshennvm
Add support for drop_last=False
#96 opened Feb 1, 2024 by odelalleau
4
Changing num_rollout_samples modifies the validation set in PPO bug Something isn't working
#90 opened Jan 23, 2024 by odelalleau
Force rampup_batch_size=None in config
#83 opened Jan 18, 2024 by shengyangs
3
Padding impacts parallel_logits computation, affecting PPO logprobs bug Something isn't working
#68 opened Dec 28, 2023 by shengyangs
1
The learning rate schedule is generally incorrect when max_steps is not set bug Something isn't working
#65 opened Dec 19, 2023 by odelalleau
Consider swapping parameters less often in DPO
#61 opened Dec 18, 2023 by odelalleau
Implement the KTO algorithm
#60 opened Dec 15, 2023 by odelalleau
1
Add the SFT tutorial
#59 opened Dec 15, 2023 by shengyangs
7
More helpful error message when failing to connect to critic server
#58 opened Dec 14, 2023 by odelalleau
GPTSFTChatDataset loss_mask becomes all False when prompt length > max_seq_length bug Something isn't working
#57 opened Dec 13, 2023 by shengyangs
Create a Template class to standardize prompt styles
#56 opened Dec 11, 2023 by shengyangs
1
code cleanup proposal
#54 opened Dec 9, 2023 by gshennvm
Support overwriting model config nodes instead of merging them
#53 opened Dec 8, 2023 by odelalleau
@gleibovich-nvidia
1
Unification of preference dataset format
#51 opened Dec 8, 2023 by odelalleau
ProTip! Updated in the last three days: updated:>2024-05-10.