Train LLM from scratch for $5 USD - Research.
Python · active 2025-12-13 → 2026-03-21 (UTC)
Activity over time
Daily event counts in the loaded window
Line chart, 99 days from 2025-12-13 to 2026-03-21. Pushes: 140 total, peak 25 in a day. Pull requests: 34 total, peak 5 in a day. Issues: 14 total, peak 4 in a day. Comments: 53 total, peak 9 in a day. Stars: 73 total, peak 20 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 |
|---|---|---|---|---|
| vukrosic | 206 | 142 | 19 | 32 |
| RohanKhanBD | 14 | 0 | 5 | 8 |
| shehab-ashraf | 5 | 0 | 1 | 4 |
| Ffinnis | 4 | 0 | 2 | 1 |
| marc-shade | 3 | 0 | 0 | 3 |
| bigwolfeman | 3 | 0 | 0 | 3 |
| github-actions[bot] | 3 | 0 | 0 | 3 |
| Tan9uy | 2 | 0 | 2 | 0 |
| VishnuTejaJ | 2 | 0 | 2 | 0 |
| toheedakhtar | 2 | 0 | 2 | 0 |
| code2tan | 1 | 0 | 0 | 0 |
| HosLak | 1 | 0 | 0 | 0 |
| Roma-Potapov | 1 | 0 | 0 | 1 |
| batuhanozkose | 1 | 0 | 1 | 0 |
Recent activity
Latest issues, pull requests and releases
- Pull request#94batuhanozkose2026-01-03 16:25
- Issue comment#89vukrosic2026-01-02 12:57Change tokenizer to mistral 7b
- Issue comment#89vukrosic2026-01-02 11:58Change tokenizer to mistral 7b
- Issue comment#89RohanKhanBD2026-01-02 10:57Change tokenizer to mistral 7b
- Issue comment#88vukrosic2026-01-02 10:22merge QKVO projections
- Pull request#88shehab-ashraf2026-01-02 09:36
- Issue#25vukrosic2025-12-27 14:59What can we implement from Karpathy's "Best ChatGPT $100 can buy"
- Issue#87vukrosic2025-12-27 14:57Does replacing some layers with linear attention (GQA or KDA) from FLA improves the model training?
- Issue comment#84vukrosic2025-12-26 07:51Implement drop-muon optimizer
- Issue comment#84Ffinnis2025-12-26 07:23Implement drop-muon optimizer
- Pull request#70vukrosic2025-12-25 11:34
- Issue comment#70vukrosic2025-12-25 11:34Fix hardcoded sequence length limitation for RoPE
- Issue comment#83vukrosic2025-12-25 08:52Fix Windows multiprocessing error in DataLoader
- Pull request#83VishnuTejaJ2025-12-25 08:52
- Issue comment#84Roma-Potapov2025-12-25 03:29Implement drop-muon optimizer
- Issue comment#84vukrosic2025-12-24 21:38Implement drop-muon optimizer
- Pull request#83VishnuTejaJ2025-12-24 14:24
- Pull request#82RohanKhanBD2025-12-24 14:14
- Issue#12vukrosic2025-12-24 09:43Token Embedding "Smearing" ⏩
- Issue#49vukrosic2025-12-24 09:41ReLU² Research
- Issue comment#49vukrosic2025-12-24 09:41ReLU² Research
- Issue comment#76vukrosic2025-12-23 18:58Token-smear
- Pull request#76RohanKhanBD2025-12-23 18:58
- Pull request#73Ffinnis2025-12-23 17:17
- Issue comment#73vukrosic2025-12-23 17:16Integrated Parallel Transformer Block +10% on training speed
Totals cover only the window loaded into ClickHouse and count events, not GitHub's lifetime totals — 73 stars here means stars gained during the window, not the repo's star count.