Skip to content

Issues: h2oai/h2ogpt

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

Clear current search query, filters, and sorts
284 Open 885 Closed
284 Open 885 Closed
Author
Filter by author
Loading
Label
Filter by label
Loading
Use alt + click/return to exclude labels
or ⇧ + click/return for logical OR
Projects
Filter by project
Loading
Milestones
Filter by milestone
Loading
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

Reveal model probability to see confidence
#6 opened Mar 28, 2023 by pseudotensor
Controlling URLs/links and other hard references so LLM can be fine-tuned on them or at least not mess them up
#7 opened Mar 28, 2023 by pseudotensor
add tensorboard/w&b/neptune tracking
#9 opened Mar 28, 2023 by pseudotensor
Ensemble multi-task LORAs
#14 opened Mar 28, 2023 by pseudotensor
consider trl-peft for RLHF
#32 opened Apr 12, 2023 by pseudotensor
See if can get 8bit 20B to run on 20GB, as advertised
#33 opened Apr 12, 2023 by arnocandel
integrate wandb and mlflow
#43 opened Apr 18, 2023 by pseudotensor
Add issue templates
#50 opened Apr 19, 2023 by ledell
@ledell
clean training data to be formatted well into paragraphs when long
#58 opened Apr 20, 2023 by pseudotensor
RuntimeError: probability tensor contains either inf, nan or element < 0
#65 opened Apr 20, 2023 by pseudotensor
add automatic push of conversations to HF data
#76 opened Apr 22, 2023 by pseudotensor
1 task
make dataset so that not truncated if train with specific cutoff_len, e.g. post-process existing dataset before trained on to deal with at training time.
#92 opened Apr 27, 2023 by pseudotensor
8-bit checkpoint saving uses same memory as 16-bit, making 8-bit useless
#95 opened Apr 28, 2023 by pseudotensor
tracking list priority/blocker Priority: issue is blocking development or release process
#97 opened Apr 29, 2023 by pseudotensor
2 of 12 tasks
recover old streaming code for num_beams>1 not supported by transformers streamer
#106 opened May 2, 2023 by pseudotensor
@pseudotensor
avoid simultaneous tokenization for multi-process training
#129 opened May 12, 2023 by arnocandel
@arnocandel
argilla 1.7.0 requires numpy<1.24.0, but you have numpy 1.24.2 which is incompatible.
#160 opened May 23, 2023 by pseudotensor
PDF makes wraps lines, then LLM does same in response helium
#264 opened Jun 9, 2023 by pseudotensor
revert when langchain keep_separator works
#273 opened Jun 10, 2023 by pseudotensor
@pseudotensor
add PDF test cases helium
#288 opened Jun 13, 2023 by pseudotensor
@pseudotensor
control max_time limits for API as well
#290 opened Jun 13, 2023 by pseudotensor
@pseudotensor
if hit stop, write uuid file that stopping.py can detect and stop, so generation really stops
#298 opened Jun 16, 2023 by pseudotensor
@pseudotensor
Extend context length during fine-tuning
#317 opened Jun 22, 2023 by arnocandel
pick 1 response out of all in multi-model view, kinda like bard
#342 opened Jun 26, 2023 by pseudotensor
@pseudotensor
drop-down pick which to use in multi model view
#343 opened Jun 26, 2023 by pseudotensor
@pseudotensor
ProTip! Type g p on any issue or pull request to go back to the pull request listing page.