用极简依赖实现高效的强化学习语言模型训练。亮点：1. 仅依赖`tokenizers`和`pytorch`，无`transformers`和`vLLM`；2. 专为单A40 GPU优化，仅需几小时即可出好结果；3. 支持Token级策略梯度损失，训练更稳定
Python · active 2025-04-13 → 2026-10-28 (UTC)
Activity over time
Daily event counts in the loaded window
Line chart, 564 days from 2025-04-13 to 2026-10-28. Pushes: 13 total, peak 3 in a day. Pull requests: 3 total, peak 2 in a day. Issues: 12 total, peak 2 in a day. Comments: 15 total, peak 3 in a day. Stars: 1,455 total, peak 425 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 |
|---|---|---|---|---|
| NTT123 | 21 | 13 | 0 | 8 |
| JJJJerry | 3 | 0 | 0 | 1 |
| je-santos | 3 | 0 | 0 | 1 |
| fan365 | 3 | 0 | 0 | 2 |
| tangjiuqi097 | 2 | 0 | 0 | 0 |
| bodsul | 2 | 0 | 2 | 0 |
| haocornell | 2 | 0 | 0 | 1 |
| KevinDayve | 2 | 0 | 0 | 1 |
| zwx8981 | 1 | 0 | 0 | 0 |
| zixiangliu960-prog | 1 | 0 | 1 | 0 |
| warmpu | 1 | 0 | 0 | 1 |
| lonelydancer | 1 | 0 | 0 | 0 |
| hkxIron | 1 | 0 | 0 | 0 |
Recent activity
Latest issues, pull requests and releases
Totals cover only the window loaded into ClickHouse and count events, not GitHub's lifetime totals — 1,455 stars here means stars gained during the window, not the repo's star count.