Sparsify transformers with SAEs and transcoders
active 2025-03-16 → 2026-07-20 (UTC)
Activity over time
Daily event counts in the loaded window
Line chart, 492 days from 2025-03-16 to 2026-07-20. Pushes: 138 total, peak 29 in a day. Pull requests: 49 total, peak 19 in a day. Issues: 19 total, peak 3 in a day. Comments: 78 total, peak 17 in a day. Stars: 89 total, peak 3 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 |
|---|---|---|---|---|
| luciaquirke | 186 | 73 | 38 | 51 |
| neverix | 31 | 30 | 1 | 0 |
| pre-commit-ci[bot] | 23 | 20 | 3 | 0 |
| norabelrose | 17 | 7 | 4 | 3 |
| kernel-loophole | 12 | 0 | 0 | 6 |
| Copilot | 11 | 0 | 0 | 3 |
| anthonyduong9 | 8 | 0 | 1 | 6 |
| github-actions[bot] | 7 | 7 | 0 | 0 |
| avgalichin | 4 | 0 | 0 | 2 |
| visionxyz | 3 | 0 | 0 | 2 |
| Abzinger | 2 | 0 | 0 | 0 |
| jacklanda | 2 | 0 | 1 | 1 |
| CLAassistant | 2 | 0 | 0 | 2 |
| chanind | 1 | 0 | 0 | 0 |
| SrGonao | 1 | 1 | 0 | 0 |
| okihnjo | 1 | 0 | 0 | 1 |
| zsquaredz | 1 | 0 | 0 | 0 |
| behroozazarkhalili | 1 | 0 | 0 | 0 |
Recent activity
Latest issues, pull requests and releases
- Issue comment#131norabelrose2026-04-13 15:19Docs: Clarify micro_acc_steps functionality #111
- Issue#132chanind2026-02-11 01:10b_dec is incorrectly added to topk aux loss
- Releasegithub-actions[bot]2025-11-17 23:06v1.3.0
- Issue comment#124jacklanda2025-11-13 08:22What is "attribution_graph" in the forked "clt-training" repo?
- Pull request#122pre-commit-ci[bot]2025-10-06 17:41
- Pull request#121luciaquirke2025-10-01 21:54
- Issue#120luciaquirke2025-10-01 04:28IndexError: Dimension out of range (expected to be in range of [-1, 0], but got 1)
- Issue#120luciaquirke2025-10-01 04:28IndexError: Dimension out of range (expected to be in range of [-1, 0], but got 1)
- Pull request#121pre-commit-ci[bot]2025-09-29 17:20
- Issue#120behroozazarkhalili2025-09-28 17:42IndexError: Dimension out of range (expected to be in range of [-1, 0], but got 1)
- Issue comment#117luciaquirke2025-09-25 06:29Evaluate SAEs with KL divergence when grafted into the model
- Pull request#117luciaquirke2025-09-25 06:29
- Releasegithub-actions[bot]2025-09-25 06:14v1.2.1
- Issue comment#108luciaquirke2025-09-25 06:09Does the code support multi node training?
- Issue#108luciaquirke2025-09-25 06:09Does the code support multi node training?
- Issue comment#116luciaquirke2025-09-23 01:50fix: auto-detect dtype from safetensors file to resolve loading mismatch
- Issue comment#116visionxyz2025-09-23 01:29fix: auto-detect dtype from safetensors file to resolve loading mismatch
- Pull request#118luciaquirke2025-09-22 23:19
- Releasegithub-actions[bot]2025-09-22 22:57v1.2.0
- Issue comment#113luciaquirke2025-09-22 22:53feat: Exclude tokens defined by user from training
- Pull request#114luciaquirke2025-09-22 22:38
- Issue comment#117luciaquirke2025-09-22 22:34Evaluate SAEs with KL divergence when grafted into the model
- Issue comment#117kernel-loophole2025-09-22 07:30Evaluate SAEs with KL divergence when grafted into the model
- Issue comment#117luciaquirke2025-09-18 11:18Evaluate SAEs with KL divergence when grafted into the model
- Issue#115visionxyz2025-09-11 05:25Problem while Running the Code in README
Totals cover only the window loaded into ClickHouse and count events, not GitHub's lifetime totals — 89 stars here means stars gained during the window, not the repo's star count.