thunlp/OPD

• highabout#1

  Update GitHub repository description for clarity

  原因：

  当前

  Rethinking On-Policy Distillation of Large Language Models: Phenomenology, Mechanism, and Recipe

  复制粘贴的修复

  复制

  Official code for 'Rethinking On-Policy Distillation of LLMs', providing techniques and insights for effective on-policy knowledge distillation.

  待办完成忽略
• highlicense#2

  Add a LICENSE file to the repository

  原因：

  复制粘贴的修复

  复制

  Create a `LICENSE` file in the repository root with a standard open-source license (e.g., MIT, Apache-2.0, GPL-3.0) that reflects the project's intended usage.

  待办完成忽略
• mediumabout#3

  Expand repository topics and add a homepage URL

  原因：

  当前

  Topics: llms, mechanism, on-policy-distillation. Homepage: (none).

  复制粘贴的修复

  复制

  Update repository settings: set topics to `llms, mechanism, on-policy-distillation, knowledge-distillation, large-language-models, deep-learning, nlp, machine-learning, reinforcement-learning`. Set the homepage URL to `https://arxiv.org/abs/2604.13016`.

  待办完成忽略

本次扫描解析到的品类 GEO 通道：google/gemini-2.5-flash, deepseek/deepseek-v4-flash

• 品类问题

  How to efficiently distill large language models to reduce size and improve inference speed?

  你：未被推荐
  AI 推荐顺序：

  1. Hugging Face Transformers
  2. Trainer
  3. DistilBERT
  4. TinyBERT
  5. MiniLM
  6. OpenVINO Toolkit
  7. ONNX Runtime
  8. NVIDIA TensorRT
  9. TextBrewer
  10. bitsandbytes
  11. AutoGPTQ
  12. GPTQ

  AI 推荐了 12 个替代方案，却始终没点名 thunlp/OPD。这就是要补上的差距。

  查看 AI 完整回答
• 品类问题

  Seeking techniques for effective on-policy knowledge distillation in large language model applications.

  你：未被推荐
  AI 推荐顺序：

  1. trl library (huggingface/trl)
  2. acme (deepmind/acme)
  3. AlphaCode
  4. Constitutional AI
  5. stable-baselines3 (DLR-RM/stable-baselines3)
  6. transformers library (huggingface/transformers)

  AI 推荐了 6 个替代方案，却始终没点名 thunlp/OPD。这就是要补上的差距。

  查看 AI 完整回答

• Metadata completeness

  warn

  建议：

• README presence

  pass

• Compared to common alternatives in this category, what is the core differentiator of thunlp/OPD?

  pass
  ✓AI 明确点名了 thunlp/OPD

  AI 的回答可能信誓旦旦却是错的。请按事实核对：技术栈、目标人群、差异化点是不是和你实际的对得上？

• If a team adopts thunlp/OPD in production, what risks or prerequisites should they evaluate first?

  pass
  ✓AI 明确点名了 thunlp/OPD

  AI 的回答可能信誓旦旦却是错的。请按事实核对：技术栈、目标人群、差异化点是不是和你实际的对得上？

• In one sentence, what problem does the repo thunlp/OPD solve, and who is the primary audience?

  pass
  ✓AI 明确点名了 thunlp/OPD

  AI 的回答可能信誓旦旦却是错的。请按事实核对：技术栈、目标人群、差异化点是不是和你实际的对得上？