{"podcast":{"title":"Daily Paper Cast","slug":"daily-paper-cast-7079649","podcast_index_feed_id":7079649,"rss_url":"https://feeds.transistor.fm/daily-paper-cast-ai","website_url":"https://dailypapercast.transistor.fm/","image_url":"https://img.transistorcdn.com/IxaBeiMluxrMS9W9wB8hFMfmvH27KvwaSMzuhucupn0/rs:fill:0:0:1/w:1400/h:1400/q:60/mb:500000/aHR0cHM6Ly9pbWct/dXBsb2FkLXByb2R1/Y3Rpb24udHJhbnNp/c3Rvci5mbS81Zjg1/YzRhODczMDU4MmE4/OGMwN2FiNDlmYzI2/MDliMi5qcGVn.jpg","author":"Jingwen Liang, Gengyu Wang","episode_count":2000,"summary":"We update every weekday to discuss highest-voted papers from Huggingface Daily Paper (https://huggingface.co/papers). Both the podcast scripts and audio are generated by AI. Feedback and suggestions are welcome! Email us: dailypapercast.ai@gmail.com Creator: Jingwen Liang, 3D ML, https://www.linkedin.com/in/jingwen-liang/ Gengyu Wang, LLM ML, http://wanggengyu.com Listen on: Spotify: https://open.spotify.com/show/21nrhmdaA8qoBiH8q03NXL Apple Podcast: https://podcasts.apple.com/us/podcast/daily-paper-cast/id1777620236 Cover Image by Kawen Kuang https://kawen.art","last_synced_at":"2026-09-09T20:18:08.781137+00:00","page_url":"https://stenobird.com/podcast/daily-paper-cast-7079649"},"episode":{"title":"Weak-to-Strong Generalization via Direct On-Policy Distillation","slug":"weak-to-strong-generalization-via-direct-on-policy-distillation","published_at":"2026-07-15T03:45:23+00:00","page_url":"https://stenobird.com/podcast/daily-paper-cast-7079649/weak-to-strong-generalization-via-direct-on-policy-distillation","show_page_url":"https://stenobird.com/podcast/daily-paper-cast-7079649","url":"https://share.transistor.fm/s/c2959ca2","audio_url":"https://media.transistor.fm/c2959ca2/e16f89b4.mp3","summary":"🤗 Upvotes: 100 | cs.LG, cs.AI, cs.CL Authors: Shiyuan Feng, Huan-ang Gao, Haohan Chi, Hanlin Wu, Zhilong Zhang, Zheng Jiang, Bingxiang He, Wei-Ying Ma, Ya-Qin Zhang, Hao Zhou Title: Weak-to-Strong Generalization via Direct On-Policy Distillation Arxiv: http://arxiv.org/abs/2607.05394v2 Abstract: Reinforcement learning with verifiable rewards (RLVR) is a powerful recipe for improving language-model reasoning, but it is expensive to repeat on every new strong model because the target model must generate many rollouts during training. As models scale, post-training itself becomes a bottleneck. We study a weak-to-strong alternative: run RL on a smaller model where rollouts are cheaper, then reuse what that RL run learned to improve a stronger target model. Directly distilling the post-RL weak teacher is not enough, because the teacher's final policy mixes useful RL gains with the limitations of the smaller model. We propose Direct On-Policy Distillation (Direct-OPD), which transfers the teacher's RL-induced policy shift instead. Direct-OPD compares the post-RL teacher with its own pre-RL reference and treats their log-ratio as a dense implicit reward for the student. In plain terms, the checkpoint pair tells us which actions RL made the weak model more or less likely to take, and Direct-OPD applies that signal on the stronger student's own on-policy states. This directly reuses the weak model's RL supervision signal without running sparse-reward RL on the target model. Empirically, Direct-OPD consistently leverages weaker teachers to improve stronger target models; notably, it boosts Qwen3-1.7B from 48.3% to 58.3% on AIME 2024 in just 4 hours on 8 A100 GPUs. It outperforms step-matched direct RL and enables the sequential composition of multiple policy shifts. Our results…","meta_description":"🤗 Upvotes: 100 | cs.LG, cs.AI, cs.CL Authors: Shiyuan Feng, Huan-ang Gao, Haohan Chi, Hanlin Wu, Zhilong Zhang, Zheng Jiang, Bingxiang He, Wei-Ying Ma, Ya…","key_points":[],"chapters":[],"topics":[],"duration_seconds":1182,"processing_state":"not_requested","actions":[{"name":"request_transcript","method":"POST","url":"https://stenobird.com/v1/public/podcasts/daily-paper-cast-7079649/episodes/weak-to-strong-generalization-via-direct-on-policy-distillation/transcription-requests","description":"Idempotently request low-priority transcript generation for this episode."},{"name":"read_markdown","method":"GET","url":"https://stenobird.com/podcast/daily-paper-cast-7079649/weak-to-strong-generalization-via-direct-on-policy-distillation.md","description":"Read the agent-friendly Markdown representation of this episode resource."}]}}