[NeurIPS D&B '25] The one-stop repository for large language model (LLM) unlearning. Supports TOFU, MUSE, WMDP, and many unlearning methods with easy feature extensibility.
active 2025-02-25 → 2026-05-23 (UTC)
Activity over time
Daily event counts in the loaded window
Line chart, 453 days from 2025-02-25 to 2026-05-23. Pushes: 25 total, peak 5 in a day. Pull requests: 78 total, peak 13 in a day. Issues: 116 total, peak 15 in a day. Comments: 370 total, peak 25 in a day. Stars: 245 total, peak 9 in a day.
- Pushes
- Pull requests
- Issues
- Comments
- Stars
Stars, PRs, issues and forks are under-captured in the later part of this window. GH Archive progressively stopped capturing non-push events during 2026 — −95% or worse by the end of the window. Every series here except Pushes fades for that reason, so a decline above reflects the archive, not this repository. Pushes stay reliable throughout, so read them, and the contributor counts derived from them, as the real signal. Data health has the measurements.
Top contributors
Pushes, PRs, issues, reviews and comments — stars and forks excluded, so this is contribution rather than popularity
| Contributor | Contributions | Pushes | PRs | Comments |
|---|---|---|---|---|
| molereddy | 212 | 16 | 23 | 128 |
| Dornavineeth | 144 | 9 | 19 | 78 |
| ruidazeng | 48 | 0 | 9 | 31 |
| ma-kjh | 24 | 0 | 2 | 14 |
| filyp | 21 | 0 | 1 | 14 |
| tahaEntesari | 12 | 0 | 3 | 6 |
| sun1187 | 12 | 0 | 0 | 6 |
| chengjiali | 11 | 0 | 2 | 6 |
| c0syfeng | 11 | 0 | 0 | 7 |
| Atry | 9 | 0 | 1 | 7 |
| sh-qiangchen | 9 | 0 | 0 | 6 |
| Copilot | 9 | 0 | 0 | 8 |
| moro0v0 | 9 | 0 | 0 | 5 |
| shariqahn | 8 | 0 | 0 | 6 |
| liao1123 | 7 | 0 | 0 | 4 |
| Puning97 | 7 | 0 | 1 | 5 |
| wzh4464 | 4 | 0 | 4 | 0 |
| dyishappy | 4 | 0 | 0 | 3 |
| yuuee-www | 4 | 0 | 0 | 3 |
| zijianh4 | 4 | 0 | 0 | 1 |
Recent activity
Latest issues, pull requests and releases
- Pull request#186wzh44642026-03-30 06:42
- Issue comment#178Dornavineeth2026-03-07 07:24PEFT LoRA transformer layers not trained when gradient_checkpointing=True — enable_input_require_grads() missing
- Pull request#175filyp2026-03-07 07:02
- Issue comment#145chengjiali2026-03-07 05:07fix: random sampling in ForgetRetainDataset
- Issue#178Anya-wUw2026-03-05 21:39PEFT LoRA transformer layers not trained when gradient_checkpointing=True — enable_input_require_grads() missing
- Issue comment#175filyp2026-03-02 19:41Upgrading transformers version to 4.51.3 to support recent models
- Issue comment#175Dornavineeth2026-03-02 03:35Upgrading transformers version to 4.51.3 to support recent models
- Issue comment#175Dornavineeth2026-02-23 02:28Upgrading transformers version to 4.51.3 to support recent models
- Issue comment#175filyp2026-02-22 10:47Upgrading transformers version to 4.51.3 to support recent models
- Issue comment#145filyp2026-02-22 09:45fix: random sampling in ForgetRetainDataset
- Pull request#176TomerG7112026-02-11 15:41
- Issue comment#172gnueaj2026-02-03 07:57Computing individual privacy score (sMIA) for MIA Metrics Question
- Pull request#174TomerG7112026-01-26 12:19
- Issue comment#155filyp2026-01-24 11:39Fix compute_loss signature compatibility with transformers >= 4.46
- Issue#172jmp-32026-01-20 18:31Computing individual privacy score (sMIA) for MIA Metrics Question
- Pull request#170wzh44642026-01-19 08:13
- Pull request#169wzh44642026-01-17 12:56
- Pull request#168wzh44642026-01-15 07:04
- Issue#143Dornavineeth2026-01-12 04:03Script for Faithfulness and Robustness (Quantization and Relearning)?
- Issue#166Dornavineeth2026-01-12 04:03Any example command or scripts for launching an unlearning process with RMU on WMDP?
- Issue#150nxZhai2026-01-11 07:53Metric aggregations and Overall aggregate score
- Issue#161Dornavineeth2026-01-11 03:20The results of the GA and NPO tests on the TOFU dataset showed almost no change.
- Issue comment#161Duguce2026-01-11 03:11The results of the GA and NPO tests on the TOFU dataset showed almost no change.
- Issue comment#143Dornavineeth2026-01-11 02:41Script for Faithfulness and Robustness (Quantization and Relearning)?
- Issue comment#144Dornavineeth2026-01-11 02:39Significant difference in results with different GPU counts on WMDP bio dataset using GradDiff
Totals cover only the window loaded into ClickHouse and count events, not GitHub's lifetime totals — 245 stars here means stars gained during the window, not the repo's star count.