Home for "How To Scale Your Model", a short blog-style textbook about scaling LLMs on TPUs. 中文翻译及学习
HTML · active 2025-04-06 → 2026-07-13 (UTC)
Activity over time
Daily event counts in the loaded window
Line chart, 464 days from 2025-04-06 to 2026-07-13. Pushes: 179 total, peak 16 in a day. Pull requests: 39 total, peak 4 in a day. Issues: 4 total, peak 1 in a day. Comments: 49 total, peak 12 in a day. Stars: 358 total, peak 28 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 |
|---|---|---|---|---|
| jacobaustin123 | 216 | 156 | 9 | 22 |
| yashk2810 | 27 | 0 | 3 | 11 |
| github-actions[bot] | 20 | 20 | 0 | 0 |
| kiankyars | 9 | 0 | 4 | 3 |
| google-cla[bot] | 5 | 0 | 0 | 5 |
| jaustin-ant | 5 | 0 | 5 | 0 |
| theahura | 2 | 0 | 0 | 1 |
| nouiz | 2 | 0 | 2 | 0 |
| tkoeppe | 2 | 0 | 1 | 1 |
| say4n | 2 | 0 | 1 | 1 |
| jimlinntu | 2 | 0 | 1 | 1 |
| stefanZorcic | 2 | 0 | 0 | 1 |
| boyuzhangus | 1 | 0 | 0 | 1 |
| LeonEricsson | 1 | 0 | 1 | 0 |
| Kriston-12 | 1 | 0 | 0 | 0 |
| nilesh-patil | 1 | 0 | 0 | 0 |
| eitanturok | 1 | 0 | 1 | 0 |
| renxida | 1 | 0 | 1 | 0 |
| Lukas-Zenick | 1 | 0 | 1 | 0 |
| malithj | 1 | 0 | 1 | 0 |
Recent activity
Latest issues, pull requests and releases
- Issue comment#32say4n2026-05-23 21:56pdf version?
- Pull request#117say4n2026-05-22 23:34
- Pull request#106jaustin-ant2026-05-09 23:23
- Issue#104Kriston-122026-03-07 06:01Architecture diagram shows legacy Post-Norm but includes modern GQA/GLU
- Issue comment#102kiankyars2026-03-05 05:57Fix notation bugs in KV cache sharding and code errors in inference roofline
- Issue comment#98boyuzhangus2026-01-26 14:50Fix “4 big VPU” phrasing in gpus.md
- Pull request#99jaustin-ant2026-01-25 20:42
- Pull request#100jaustin-ant2026-01-25 20:42
- Pull request#99jaustin-ant2026-01-23 20:10
- Pull request#97jimlinntu2026-01-20 02:26
- Pull request#96jaustin-ant2026-01-19 21:27
- Issue comment#95jacobaustin1232025-12-14 20:13Clarify Problem 5 to specify H100 SXM and bfloat16
- Pull request#95renxida2025-12-14 20:09
- Issue#91jacobaustin1232025-12-12 03:08Question: Any plans for multi-language support for the Scaling Book?
- Pull request#93lk-chen2025-12-12 03:06
- Pull request#92tkoeppe2025-11-27 17:29
- Issue comment#79levskaya2025-10-25 05:24Updates JAX automatic parallelism section to reduce the complexity of explicit mode sharding.
- Issue comment#87jacobaustin1232025-10-20 15:44Fix tiling FLOP calculation.
- Pull request#86malithj2025-10-19 20:08
- Pull request#84jacobaustin1232025-10-06 14:16
- Issue comment#85theahura2025-10-05 06:36Chapter 3, Question 2: AllGather Latency Derivation
- Issue#85theahura2025-10-05 06:36Chapter 3, Question 2: AllGather Latency Derivation
- Issue comment#84google-cla[bot]2025-10-05 02:09Fix some small grammar in the sharding.md section
- Pull request#83jacobaustin1232025-09-28 22:38
- Pull request#81fedelebron2025-09-21 16:09
Totals cover only the window loaded into ClickHouse and count events, not GitHub's lifetime totals — 358 stars here means stars gained during the window, not the repo's star count.