When FP8 Training Stops Punishing Bad Answers

Featured

• OmniEdu: Open Foundation Models for Learning and Teaching score 10
  入选 HF Daily Papers; HF 热度: 66 upvotes (+4); 有代码实现; 关键词(2): fine-tune, reasoning
• UltraTex: Unleashing 2K Multi-View Diffusion for 3D Texturing score 6
  入选 HF Daily Papers; 有代码实现; 关键词(3): scaling, production, latency
• Towards Full Pipeline FP8 Reinforcement Learning for LLMs score 5
  入选 HF Daily Papers; HF 热度: 4 upvotes (+1); 关键词(5): scaling, quantization, GRPO, agentic, reasoning

Also Worth Noting

• OptiSkill: A Hierarchical and Evolving SkillBank for LLM-Based Optimization Modeling score 4
  关键词(1): agentic; 顶会接收: EMNLP
• Inherit4Rec: Parameter Inheritance for Efficient Scaling of Recommendation Models score 4
  机构: Huawei; 关键词(2): scaling, serving
• Beyond Average Error through Oracle-Informed Stress Tests for Time-Series Forecasting score 3
  机构: Tsinghua
• Rethinking Pivot Programming Languages in Code Language Models score 3
  顶会接收: EMNLP
• MolSC: Leveraging Substituent Contributions to Enhance Fine-grained Molecular Understanding in LLMs score 3
  顶会接收: EMNLP