Xiao Liu

Xiao Liu on Hugging Face Daily Papers: 17 papers, 4 in the top 3 of their day, 391 upvotes.

  1. Improving Data and Reward Design for Scientific Reasoning in Large Language Models 39 upvotes, #13 of 2026-02-10
  2. MSign: An Optimizer Preventing Training Instability in Large Language Models via Stable Rank Restoration 32 upvotes, #7 of 2026-02-09
  3. Learning from the Best, Differently: A Diversity-Driven Rethinking on Data Selection 3 upvotes, #30 of 2025-10-23
  4. Recycling Pretrained Checkpoints: Orthogonal Growth of Mixture-of-Experts for Efficient Large Language Model Pre-Training 5 upvotes, #36 of 2025-10-10
  5. Behind RoPE: How Does Causal Mask Encode Positional Information? 6 upvotes, #22 of 2025-09-26
  6. Data Mixing Agent: Learning to Re-weight Domains for Continual Pre-training 3 upvotes, #24 of 2025-07-22
  7. TL;DR: Too Long, Do Re-weighting for Effcient LLM Reasoning Compression 3 upvotes, #35 of 2025-06-04
  8. How does Alignment Enhance LLMs' Multilingual Capabilities? A Language Neurons Perspective 17 upvotes, #25 of 2025-05-28
  9. Overcoming Vocabulary Mismatch: Vocabulary-agnostic Teacher Guided Language Modeling 2 upvotes, #28 of 2025-03-26
  10. Process-based Self-Rewarding Language Models 35 upvotes, #3 of 2025-03-06
  11. Optimizing Large Language Model Training Using FP4 Quantization 32 upvotes, #2 of 2025-01-29
  12. Sigma: Differential Rescaling of Query, Key and Value for Efficient Language Models 40 upvotes, #3 of 2025-01-24
  13. EpiCoder: Encompassing Diversity and Complexity in Code Generation 9 upvotes, #13 of 2025-01-09
  14. Ensembling Large Language Models with Process Reward-Guided Tree Search for Better Complex Reasoning 12 upvotes, #7 of 2024-12-25
  15. DA-Code: Agent Data Science Code Generation Benchmark for Large Language Models 4 upvotes, #15 of 2024-10-14
  16. Rho-1: Not All Tokens Are What You Need 72 upvotes, #1 of 2024-04-12
  17. AR-Diffusion: Auto-Regressive Diffusion Model for Text Generation 3 upvotes, #6 of 2023-05-17

Data: hysts-bot-data/daily-papers-stats and the Daily Papers API. Open data: tardellirs/paper-pulse-data. Sister project: Model Pulse, the download history of every model on the Hub.