Lune

Profiles2024-2026

Hao Peng

University of Illinois Urbana -Champaign

14Papers
10Last author
26Unique co-authors
0First author
0Solo

First and last author come from byline position, which some venues assign alphabetically.

Recent papers14

  1. Good SFT Optimizes for SFT, Better SFT Prepares for Reinforcement LearningAuthor 5 of 5ICML 2026 · 8 citations
  2. The Unreasonable Effectiveness of Entropy Minimization in LLM ReasoningAuthor 5 of 5NeurIPS 2025 · 185 citations
  3. The Best Instruction-Tuning Data are Those That FitAuthor 3 of 3NeurIPS 2025 · 59 citations
  4. Reinforcement Learning Finetunes Small Subnetworks in Large Language ModelsAuthor 4 of 4NeurIPS 2025 · 43 citations
  5. Context Length Alone Hurts LLM Performance Despite Perfect RetrievalAuthor 10 of 10EMNLP Findings 2025 · 38 citations
  6. FACTCHECKMATE: Preemptively Detecting and Mitigating Hallucinations in LMsAuthor 4 of 4EMNLP Findings 2025 · 3 citations
  7. Improving Influence-based Instruction Tuning Data Selection for Balanced Learning of Diverse CapabilitiesAuthor 4 of 4EMNLP Findings 2025 · 1 citation
  8. A Little Goes a Long Way: Efficient Long Context Training and Inference with Partial ContextsAuthor 5 of 5ICLR 2025
  9. Retrieval Head Mechanistically Explains Long-Context FactualityAuthor 4 of 5ICLR 2025
  10. MINT: Evaluating LLMs in Multi-turn Interaction with Tools and Language FeedbackAuthor 6 of 7ICLR 2024 · 308 citations
  11. Data Engineering for Scaling Language Models to 128K ContextAuthor 7 of 7ICML 2024 · 204 citations
  12. CRAFT: Customizing LLMs by Creating and Retrieving from Specialized ToolsetsAuthor 5 of 6ICLR 2024 · 117 citations
  13. TRAM: Bridging Trust Regions and Sharpness Aware MinimizationAuthor 4 of 4ICLR 2024 · 6 citations
  14. ActionIE: Action Extraction from Scientific Literature with Programming LanguagesAuthor 7 of 9ACL 2024
Dusk over the sea between two cliffs drawn in fine vertical lines