|
| 1 | +# Weekly Content Update Task |
| 2 | + |
| 3 | +You are updating an "awesome list" of AI-driven automated model training tools. |
| 4 | + |
| 5 | +## Your Goal |
| 6 | + |
| 7 | +Search for NEW tools, frameworks, and projects related to algorithm automation and model training that have been released or significantly updated in the past 1-2 weeks. Update both `README.md` (English) and `算法自动化训练.md` (Chinese) with any new findings. |
| 8 | + |
| 9 | +## Search Scope |
| 10 | + |
| 11 | +Search across these categories (use WebSearch for each): |
| 12 | + |
| 13 | +1. **Autonomous ML experiment / research frameworks** - new AI agent systems that run experiments autonomously |
| 14 | +2. **RL alignment training** (RLHF/GRPO/DPO) - new frameworks, algorithms, or significant updates |
| 15 | +3. **LLM fine-tuning frameworks** - new tools or major releases (Unsloth, Axolotl, LlamaFactory, etc.) |
| 16 | +4. **Synthetic data generation** - new tools for generating training data at scale |
| 17 | +5. **Inference engines** - new or updated engines relevant to RL training loops |
| 18 | +6. **Coding agents** - new open-source agents for writing training scripts |
| 19 | +7. **Model merging & quantization** - new compression techniques or tools |
| 20 | +8. **Multimodal training** - new vision-language or audio-language training frameworks |
| 21 | +9. **Evaluation & benchmarks** - new LLM/agent benchmarks |
| 22 | +10. **Experiment tracking & MLOps** - new tools or significant updates |
| 23 | + |
| 24 | +## Quality Criteria |
| 25 | + |
| 26 | +Only add tools that meet ALL of these: |
| 27 | +- **Directly usable** for automated model training workflows |
| 28 | +- **Open source** with active maintenance (prefer projects with 100+ GitHub stars or from reputable labs) |
| 29 | +- **Genuinely new** - not already listed in the existing files |
| 30 | +- **Significant** - not minor forks or trivial wrappers |
| 31 | + |
| 32 | +## Update Process |
| 33 | + |
| 34 | +1. Read both `README.md` and `算法自动化训练.md` to understand the current content |
| 35 | +2. Search for new developments using WebSearch (at least 6 targeted searches) |
| 36 | +3. For each new tool found, verify it meets quality criteria |
| 37 | +4. Add entries to BOTH files in the appropriate sections: |
| 38 | + - README.md: add table rows matching the existing format `| [Name](url) | Description | Key Highlight |` |
| 39 | + - 算法自动化训练.md: add detailed entries matching the existing format with GitHub link, description, key features, and use cases |
| 40 | +5. If you find new trends, update the Trends section in both files |
| 41 | +6. Update the date in both files to today's date |
| 42 | +7. If no significant new tools are found, create a file `.github/last-update-log.md` noting "No new tools found on [date]" and do NOT modify the main files |
| 43 | + |
| 44 | +## Important Rules |
| 45 | + |
| 46 | +- Do NOT remove or modify existing entries |
| 47 | +- Do NOT change the structure or formatting conventions |
| 48 | +- Maintain consistent numbering in the Chinese doc |
| 49 | +- Keep table column counts consistent in README.md |
| 50 | +- Write Chinese descriptions for 算法自动化训练.md, English for README.md |
0 commit comments