RLVR训练数据不够用？从互联网文本里无限合成就行

重点关注

• PaperBanana: Automating Academic Illustration for AI Scientists score 12
  机构: Yale；入选 HF Daily Papers；HF 热度: 137 upvotes (+4)；关键词(2): agents, agentic
• Golden Goose: A Simple Trick to Synthesize Unlimited RLVR Tasks from Unverifiable Internet Text score 9
  入选 HF Daily Papers；HF 热度: 83 upvotes (+4)；关键词(4): scaling, post-training, pre-training, reasoning
• VoxServe: Streaming-Centric Serving System for Speech Language Models score 7
  入选 HF Daily Papers；HF 热度: 6 upvotes (+2)；关键词(6): efficiency, serving, inference, latency, throughput
• Quartet II: Accurate LLM Pre-Training in NVFP4 by Improved Unbiased Gradient Estimation score 9
  入选 HF Daily Papers；HF 热度: 53 upvotes (+4)；关键词(2): quantization, pre-training
• THINKSAFE: Self-Generated Safety Alignment for Reasoning Models score 9
  入选 HF Daily Papers；HF 热度: 38 upvotes (+4)；关键词(8): lightweight, distillation, fine-tuning, GRPO, alignment
• ReGuLaR: Variational Latent Reasoning Guided by Rendered Chain-of-Thought score 9
  入选 HF Daily Papers；HF 热度: 34 upvotes (+4)；关键词(4): efficient, efficiency, compression, reasoning
• Residual Context Diffusion Language Models score 9
  入选 HF Daily Papers；HF 热度: 30 upvotes (+4)；关键词(2): diffusion, reasoning
• Statistical Estimation of Adversarial Risk in Large Language Models under Best-of-N Sampling score 9
  入选 HF Daily Papers；HF 热度: 20 upvotes (+4)；关键词(5): scaling, evaluation, cost, safety, jailbreak
• FourierSampler: Unlocking Non-Autoregressive Potential in Diffusion Language Models via Frequency-Guided Generation score 9
  入选 HF Daily Papers；HF 热度: 20 upvotes (+4)；关键词(2): inference, diffusion
• TTCS: Test-Time Curriculum Synthesis for Self-Evolving score 8
  入选 HF Daily Papers；HF 热度: 32 upvotes (+4)；关键词(1): reasoning

也值得关注

• AI-Enabled Waste Classification as a Data-Driven Decision Support Tool for Circular Economy and Urban Sustainability score 2
  关键词(2): efficient, real-time
• Dynamic Welfare-Maximizing Pooled Testing score 2
  关键词(2): efficient, cost
• The Blessing of Dimensionality in LLM Fine-tuning: A Variance-Curvature Perspective score 2
  关键词(3): fine-tuning, fine-tune, GRPO
• Uncertainty-Aware Multimodal Learning via Conformal Shapley Intervals score 2
  关键词(2): inference, multimodal
• ReNCE: Learning to Reason by Noise Contrastive Estimation score 2
  关键词(3): DPO, GRPO, reasoning
• Large Language Model Agents Are Not Always Faithful Self-Evolvers score 2
  关键词(2): agent, agents
• Towards Resiliency in Large Language Model Serving with KevlarFlow score 2
  关键词(4): serving, latency, throughput, evaluation
• AI and My Values: User Perceptions of LLMs' Ability to Extract, Embody, and Explain Human Values from Casual Conversations score 2
  关键词(3): alignment, agents, reasoning
• Automating Forecasting Question Generation and Resolution for AI Evaluation score 2
  关键词(2): agents, evaluation
• High-Definition 5MP Stereo Vision Sensing for Robotics score 2
  关键词(2): real-time, robotics
• Anytime Safe PAC Efficient Reasoning score 2
  关键词(5): efficient, efficiency, latency, reasoning, safety
• HeaPA: Difficulty-Aware Heap Sampling and On-Policy Query Augmentation for LLM Reinforcement Learning score 2
  关键词(5): efficiency, lightweight, latency, reasoning, cost
• Tuning the Implicit Regularizer of Masked Diffusion Language Models: Enhancing Generalization via Insights from $k$-Parity score 2
  关键词(3): fine-tuning, pre-training, diffusion