🚀 Pytorch Distributed native training library for LLMs/VLMs with OOTB Hugging Face support
Python · active 2025-07-03 → 2026-08-10 (UTC)
Activity over time
Daily event counts in the loaded window
Line chart, 404 days from 2025-07-03 to 2026-08-10. Pushes: 6,673 total, peak 89 in a day. Pull requests: 1,041 total, peak 17 in a day. Issues: 486 total, peak 26 in a day. Comments: 2,977 total, peak 54 in a day. Stars: 152 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 |
|---|---|---|---|---|
| akoumpa | 3.1K | 1.8K | 225 | 699 |
| copy-pr-bot[bot] | 2.3K | 1.8K | 0 | 440 |
| HuiyingLi | 1.1K | 680 | 84 | 253 |
| adil-a | 852 | 475 | 71 | 186 |
| hemildesai | 637 | 344 | 57 | 176 |
| thomasdhc | 539 | 295 | 89 | 83 |
| jgerh | 511 | 11 | 0 | 455 |
| chtruong814 | 259 | 74 | 130 | 22 |
| github-actions[bot] | 248 | 52 | 63 | 95 |
| yuhezhang-ai | 184 | 161 | 4 | 13 |
| ZhiyuLi-Nvidia | 153 | 74 | 16 | 46 |
| svcnvidia-nemo-ci | 149 | 18 | 92 | 24 |
| claude[bot] | 113 | 0 | 0 | 55 |
| pthombre | 107 | 79 | 10 | 13 |
| ko3n1g | 103 | 61 | 12 | 8 |
| athitten | 93 | 57 | 8 | 17 |
| BoxiangW | 82 | 38 | 10 | 26 |
| snowmanwwg | 74 | 4 | 11 | 7 |
| ffrujeri | 62 | 40 | 5 | 12 |
| torsli | 57 | 19 | 6 | 15 |
Recent activity
Latest issues, pull requests and releases
- Issue comment#3267khazic2026-08-10 06:42fix(speculative): align the GLM-5.2 DSpark draft RoPE with its target
- Issue comment#3414yuhezhang-ai2026-08-07 19:28fix(vlm): correct CP gradients and vision sharding
- Issue comment#3437yuhezhang-ai2026-08-07 19:28fix(checkpoint): save ranked RNG state per global rank
- Issue comment#3439yuhezhang-ai2026-08-07 19:11fix(peft): enable v5 expert adapters for Nemotron and MiniMax
- Issue comment#3460svcnvidia-nemo-ci2026-08-07 19:07cp: `test(checkpoint): stabilize MoE checkpoint parity gates (3440)` into `r0.6.0`
- Pull request#3460svcnvidia-nemo-ci2026-08-07 19:07
- Pull request#3440yuhezhang-ai2026-08-07 19:07
- Issue comment#3459copy-pr-bot[bot]2026-08-07 18:54cp: `fix: normalize MoE auxiliary loss during gradient accumulation (3359)` into `r0.6.0`
- Pull request#3459svcnvidia-nemo-ci2026-08-07 18:54
- Issue comment#3456athitten2026-08-07 17:34fix(gemma4): exact router scalar and fp32 reference routing for MoE parity
- Issue comment#3456athitten2026-08-07 17:32fix(gemma4): exact router scalar and fp32 reference routing for MoE parity
- Issue#3455zyzhou52026-08-07 16:50nemotron_labs_diffusion_sft.yaml nccl + cpuoffload
- Issue comment#3260HuiyingLi2026-08-01 22:09fix(checkpoint): restore Gemma4 Unified HF export keys
- Issue comment#3107saisharan01032026-08-01 09:51refactor(packing): replace global HF monkeypatch with typed metadata
- Issue comment#3347svcnvidia-nemo-ci2026-07-31 21:22cp: `fix(distributed): reuse default group for world-sized meshes (3319)` into `r0.6.0`
- Issue comment#2872yuhezhang-ai2026-07-27 20:40feat(retrieval): add optimized Nemotron VL custom path
- Issue comment#3204akoumpa2026-07-23 15:03build: cache CUDA extension source builds
- Issue comment#2987yuhezhang-ai2026-07-22 19:11feat(distributed): shape TPLinear/LinearLoRA graphs for async-TP fusion
- Issue comment#3130HuiyingLi2026-07-22 06:11feat(checkpoint): add DCP CPU offload option
- Issue comment#3110svcnemo-autobot2026-07-16 23:42perf(bench): AUT-889 use fused Adam in glm_4.5_air_te_deepep benchmark
- Pull request#3096github-actions[bot]2026-07-16 06:29
- Issue comment#3058thomasdhc2026-07-15 17:57feat: add Ministral3 embedding distillation recipe
- Issue comment#3065HuiyingLi2026-07-15 00:37fix: Kimi K2 config loading without remote code
- Issue comment#3036svcnemo-autobot2026-07-11 16:36AUT-781 In FW-CI-Templates we recently replaced tags with SHAs. Do the same for
- Issue comment#3028HuiyingLi2026-07-11 06:33fix(models): temporarily disable TE fused RoPE globally
Totals cover only the window loaded into ClickHouse and count events, not GitHub's lifetime totals — 152 stars here means stars gained during the window, not the repo's star count.