Fine-tuning & Reinforcement Learning for LLMs. š¦„ Train Qwen3, Llama 4, DeepSeek-R1, Gemma 3, TTS 2x faster with 70% less VRAM.
active 2025-08-01 ā 2026-01-02 (UTC)
Activity over time
Daily event counts in the loaded window
Line chart, 155 days from 2025-08-01 to 2026-01-02. Pushes: 32 total, peak 13 in a day. Pull requests: 16 total, peak 8 in a day. Issues: 14 total, peak 8 in a day. Comments: 17 total, peak 9 in a day. Stars: 9 total, peak 6 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 ā ā95% 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 ā stars and forks excluded, so this is contribution rather than popularity
| Contributor | Contributions | Pushes | PRs | Comments |
|---|---|---|---|---|
| OEvortex | 33 | 12 | 6 | 2 |
| Copilot | 29 | 18 | 10 | 1 |
| codeant-ai[bot] | 14 | 0 | 0 | 14 |
| danielhanchen | 2 | 2 | 0 | 0 |
| VarunGuptaPy | 1 | 0 | 0 | 0 |
Recent activity
Latest issues, pull requests and releases
- Pull request#27Copilot2025-12-10 13:17
- Issue comment#25codeant-ai[bot]2025-08-03 10:08[WIP] torchrun --nproc_per_node=4 train.py W0803 10:05:46.985000 22042 torch/distributed/run.py:766] W0803 10:05:46.985000 22042 torch/distributed/run.py:766] ***************************************** W0803 10:05:46.985000 22042 torch/distributed/run.py:766...
- Issue comment#25codeant-ai[bot]2025-08-03 10:06[WIP] torchrun --nproc_per_node=4 train.py W0803 10:05:46.985000 22042 torch/distributed/run.py:766] W0803 10:05:46.985000 22042 torch/distributed/run.py:766] ***************************************** W0803 10:05:46.985000 22042 torch/distributed/run.py:766...
- Pull request#25Copilot2025-08-03 10:06
- Pull request#23OEvortex2025-08-03 09:48
- Issue comment#23codeant-ai[bot]2025-08-03 09:40[WIP] Unsloth: Warning - Could not find DDP-wrapped model for static graph optimization Unsloth: If you encounter 'parameter marked ready twice' or 'expect_autograd_hooks_' errors, this is the likely cause GPU = Tesla V100-SXM2-16GB. Max memory = 15.766 GB. ...
- Issue comment#22codeant-ai[bot]2025-08-02 16:17[WIP] I am finetuning Qwen3-30b on 8xH100 80GB vram variant. The model should ideally take 48GB of VRAM in each GPU in DDP. But it is running out of memory with following error when I am running the training file with command :- torchrun --nproc_per_node=8 t...
- Pull request#22Copilot2025-08-02 16:16
- Issue comment#21Copilot2025-08-02 13:32[Bug] DDP uses 2x more vram
- Issue#21OEvortex2025-08-02 13:32[Bug] DDP uses 2x more vram
- Issue#20VarunGuptaPy2025-08-02 13:25[Bug] DDP is 2x vram
- Pull request#19OEvortex2025-08-02 09:08
- Issue#18OEvortex2025-08-02 09:08[Bug] Please fill in your issue title here.
- Issue comment#19codeant-ai[bot]2025-08-02 08:46[WIP] [Bug] Please fill in your issue title here.
- Pull request#19Copilot2025-08-02 08:46
- Issue#18OEvortex2025-08-02 08:46[Bug] Please fill in your issue title here.
- Issue#16OEvortex2025-08-02 04:10ddp issue
- Pull request#17OEvortex2025-08-02 04:09
- Issue comment#17codeant-ai[bot]2025-08-02 03:55[WIP] ddp issue
- Issue comment#17codeant-ai[bot]2025-08-02 03:54[WIP] ddp issue
- Pull request#17Copilot2025-08-02 03:54
- Issue#16OEvortex2025-08-02 03:54ddp issue
- Issue#14OEvortex2025-08-01 16:04[Bug] Please fill in your issue title here.
- Issue#12OEvortex2025-08-01 14:01[Bug] Please fill in your issue title here.
- Pull request#13OEvortex2025-08-01 14:01
Totals cover only the window loaded into ClickHouse and count events, not GitHub's lifetime totals ā 9 stars here means stars gained during the window, not the repo's star count.