torchax is a PyTorch frontend for JAX. It gives JAX the ability to author JAX programs using familiar PyTorch syntax. It also provides JAX-Pytorch interoperability, meaning, one can mix JAX & Pytorch syntax together when authoring ML programs, and run it in every hardware JAX can run.
active 2025-10-03 → 2026-08-07 (UTC)
Activity over time
Daily event counts in the loaded window
Line chart, 309 days from 2025-10-03 to 2026-08-07. Pushes: 66 total, peak 7 in a day. Pull requests: 30 total, peak 3 in a day. Issues: 14 total, peak 2 in a day. Comments: 38 total, peak 4 in a day. Stars: 76 total, peak 13 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 |
|---|---|---|---|---|
| qihqi | 65 | 35 | 8 | 13 |
| wdhongtw | 31 | 10 | 7 | 7 |
| ShitongWang | 12 | 11 | 1 | 0 |
| jialei777 | 11 | 6 | 3 | 1 |
| Artur-Galstyan | 8 | 0 | 2 | 3 |
| mozuysal-hubx | 6 | 0 | 0 | 4 |
| QiliangCui | 4 | 1 | 2 | 0 |
| yuyanpeng-google | 3 | 1 | 0 | 1 |
| lsy323 | 3 | 0 | 0 | 0 |
| google-cla[bot] | 3 | 0 | 0 | 3 |
| kyuyeunk | 3 | 0 | 2 | 1 |
| junjieqian | 2 | 1 | 1 | 0 |
| BioGeek | 2 | 0 | 0 | 1 |
| hfan | 2 | 0 | 0 | 0 |
| joey00072 | 2 | 0 | 0 | 2 |
| vlad-karp | 2 | 1 | 0 | 0 |
| rdyro | 1 | 0 | 0 | 1 |
| samos123 | 1 | 0 | 0 | 0 |
| salmanmkc | 1 | 0 | 1 | 0 |
| carlesoctav | 1 | 0 | 0 | 0 |
Recent activity
Latest issues, pull requests and releases
- Pull request#90wdhongtw2026-04-12 13:57
- Issue comment#85qihqi2026-04-11 00:30add rng option to jittablemodule
- Issue comment#67qihqi2026-04-02 01:39train_llama_torchtitan example: Multiple issues running on TPU v4-32
- Issue#39qihqi2026-03-18 03:13SDPA dispatch to TPU Pallas kernel is missing inverse sqrt scaling and overrides casual as always true
- Pull request#78wdhongtw2026-03-18 03:13
- Issue#82hjajoo-g2026-03-12 15:25torch.nn.utils.rnn.pad_sequence replaces entire output tensor with padding_value on TPU
- Issue comment#80wdhongtw2026-03-04 07:15Upgrade GitHub Actions for Node 24 compatibility
- Pull request#76wdhongtw2026-03-04 06:14
- Pull request#80salmanmkc2026-02-22 18:45
- Pull request#79jrplatin2026-02-22 06:23
- Pull request#76wdhongtw2026-02-20 13:08
- Pull request#69qihqi2026-01-30 06:54
- Issue comment#74google-cla[bot]2026-01-29 07:38Bump version to 0.0.12.
- Pull request#74QiliangCui2026-01-29 07:38
- Pull request#74QiliangCui2026-01-29 07:38
- Pull request#71wdhongtw2026-01-28 01:49
- Issue#68carlesoctav2026-01-22 05:47How torchax handle tied_embedding?
- Issue#67KarthikRevanuru2026-01-21 14:52train_llama_torchtitan example: Multiple issues running on TPU v4-32
- Pull request#66jack85582026-01-20 23:59
- Issue#65samos1232026-01-14 21:32Convolution backward op missing
- Pull request#63wdhongtw2025-12-22 17:28
- Issue comment#62qihqi2025-12-19 05:19Unable to export quantized models to stablehlo
- Issue comment#61qihqi2025-12-19 05:09Remove option for using flash attention
- Issue comment#61jialei7772025-12-15 17:41Remove option for using flash attention
- Issue comment#60qihqi2025-12-14 18:48Solve deprecation warning since jax 0.8.0
Totals cover only the window loaded into ClickHouse and count events, not GitHub's lifetime totals — 76 stars here means stars gained during the window, not the repo's star count.