← authors · overview

• Consmax: Hardware-friendly Alternative Softmax With Learnable Parameters
  2024 · 17 citations
• Junk DNA Hypothesis: Pruning Small Pre-trained Weights Irreversibly And Monotonically Impairs "difficult" Downstream Tasks In Llms
  2023 · 13 citations
• Achieving Personalized Federated Learning with Sparse Local Models
  2022 · 13 citations
• Demystifying the Roles of LLM Layers in Retrieval, Knowledge, and Reasoning
  2025 · 10 citations
• Are Large Kernels Better Teachers than Transformers for ConvNets?
  2023 · 1 citations
• ISO: An RLVR-Native Optimization Stack
  2026
• ISO: An RLVR-Native Optimization Stack
  2026
• Learning from the Self-future: On-policy Self-distillation for dLLMs
  2026
• Acttail: Global Activation Sparsity In Large Language Models
  2026
• When Fewer Layers Break More Chains: Layer Pruning Harms Test-Time Scaling in LLMs
  2025
• O1-Pruner: Length-Harmonizing Fine-Tuning for O1-Like Reasoning Pruning
  2025
• Mask-Enhanced Autoregressive Prediction: Pay Less Attention to Learn More
  2025
• Stable-SPAM: How to Train in 4-Bit More Stably than 16-Bit Adam
  2025
• SoS1: O1 and R1-Like Reasoning LLMs are Sum-of-Square Solvers
  2025
• LIFT the Veil for the Truth: Principal Weights Emerge after Rank Reduction for Reasoning-Focused Supervised Fine-Tuning
  2025