Skip to content

Issues: instructlab/training

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
61 Open 103 Closed
61 Open 103 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

Refactor: checkpoint, stopping logic
#178 opened Aug 26, 2024 by RobotSail updated Aug 26, 2024
Potential masking issue in get_masked_and_orig_text function
#184 opened Aug 27, 2024 by aldopareja updated Aug 27, 2024
@Maxusmusti
[FSDP] Allow training to work with FSDP in ZeRO stage-3
#201 opened Sep 12, 2024 by RobotSail updated Sep 12, 2024
@JamesKunstle
Re-introduce gradnorm/weightnorm logs
#223 opened Sep 25, 2024 by Maxusmusti updated Sep 25, 2024
Assess the performance / memory trade-off
#228 opened Sep 26, 2024 by Maxusmusti updated Sep 26, 2024
Speed up training library loads enhancement New feature or request good first issue Good for newcomers
#115 opened Jun 28, 2024 by RobotSail updated Oct 2, 2024
1
Add Support for Early Stopping Criteria enhancement New feature or request
#245 opened Oct 2, 2024 by Maxusmusti updated Oct 2, 2024
Enforce the types that are passed in from the CLI in main_ds and data_process enhancement New feature or request good first issue Good for newcomers
#27 opened Jun 17, 2024 by RobotSail updated Oct 2, 2024
1
Decrease number of required fields in TrainingArgs and TorchrunArgs
#252 opened Oct 7, 2024 by RobotSail updated Oct 7, 2024
[Feature] Support saving LoRA adapters with accelerator.save_state
#253 opened Oct 7, 2024 by JamesKunstle updated Oct 7, 2024
Division by zero error when calling find_packing_max_batch_len_and_grad_accum
#254 opened Oct 7, 2024 by RobotSail updated Oct 7, 2024
1
Bump min accelerate version to 1.0.0
#266 opened Oct 11, 2024 by RobotSail updated Oct 11, 2024
Make DeepSpeed an optional requirement
#250 opened Oct 7, 2024 by RobotSail updated Oct 13, 2024
1
Load from accelerate state fails when there's nothing to load from
#272 opened Oct 15, 2024 by RobotSail updated Oct 15, 2024
Training should avoid prints and return status to the caller
#281 opened Oct 17, 2024 by danmcp updated Oct 21, 2024
2
Add Dolomite test to smoketests
#297 opened Oct 23, 2024 by JamesKunstle updated Oct 23, 2024
Add named optional parameters to smoketest.sh
#300 opened Oct 23, 2024 by JamesKunstle updated Oct 23, 2024
Test save_samples path in smoketest
#303 opened Oct 24, 2024 by JamesKunstle updated Oct 24, 2024
Make tqdm output work in a more friendly way
#306 opened Oct 25, 2024 by RobotSail updated Oct 25, 2024
Model/Optimizer Setup functions need moving and typehints good first issue Good for newcomers
#225 opened Sep 25, 2024 by Maxusmusti updated Oct 25, 2024
1
@malinjawi
7
Allow FSDP prefetch to be configurable
#307 opened Oct 25, 2024 by RobotSail updated Oct 25, 2024
Create a changelog.md file documentation Improvements or additions to documentation
#209 opened Sep 13, 2024 by RobotSail updated Oct 30, 2024
1
Repo needs release-strategy.md document documentation Improvements or additions to documentation
#316 opened Nov 1, 2024 by nathan-weinberg updated Nov 1, 2024
Failing to run library on Kaggle
#317 opened Nov 4, 2024 by kittycattoys updated Nov 4, 2024
ilab train - stopped but not seeing error and cause
#327 opened Nov 10, 2024 by acsankar updated Nov 10, 2024
ProTip! Type g i on any issue or pull request to go back to the issue listing page.