khtsly

khtsly's activity

• All
• Models
• Datasets
• Spaces
• Buckets
• Papers
• Collections
• Community
• Posts
• Upvotes
• Likes
• Articles

upvoted a paper about 5 hours ago

FLEET: From Logits Entropy to Enhanced Trajectories in Text Generation

Paper • 2609.27657 • Published 5 days ago • 5

upvoted a paper about 24 hours ago

Rufus-Air: An Open LLM Post-Training Recipe

Paper • 2609.29421 • Published 4 days ago • 19

upvoted a paper 1 day ago

Your Transformer Can Hold Two Thoughts at Once: Evidence of Linear Superposition in LLMs

Paper • 2609.29845 • Published 4 days ago • 77

upvoted a paper 3 days ago

When2Think: Learning Difficulty-Aware Length Control for Efficient Hybrid Reasoning Models

Paper • 2609.19671 • Published 11 days ago • 54

upvoted 2 papers 5 days ago

Bellman Policy Optimization

Paper • 2609.15987 • Published 14 days ago • 48

Flash-dLLM: IO-Aware KV Caching and Parallel Decoding for Fast, Memory-Efficient Diffusion LLMs

Paper • 2609.26796 • Published 6 days ago • 36

upvoted 3 papers 10 days ago

When EOS Tokens Disagree: Understanding Length Inflation in On-Policy Distillation

Paper • 2609.20511 • Published 11 days ago • 110

DeepSeek-V4.1-Flash: Pushing the Limits of KV Cache Compression

Paper • 2609.19969 • Published 11 days ago • 186

JEPA-Anything: Learning Predictive Models across Different Worlds

Paper • 2609.20800 • Published 11 days ago • 74

upvoted 3 papers 11 days ago

VC-Attention: Value Smoothing and Softmax Casting for Low-bit Attention

Paper • 2609.15810 • Published 14 days ago • 51

ProgramDistill: From Interactive Web Apps to Verifiable Reference-Guided SWE Tasks

Paper • 2609.18805 • Published 12 days ago • 66

Rethinking Critic Learning in PPO: Understanding and Mitigating Value Flattening

Paper • 2609.18708 • Published 12 days ago • 82

upvoted 4 papers 13 days ago

SAS: Simple Attention Sparsification via End-to-End Optimization of Context Ranking

Paper • 2609.13141 • Published 17 days ago • 65

Grouped Value Attention: Efficient KV Caching via On-Demand Key Reconstruction

Paper • 2609.13285 • Published 20 days ago • 81

Atria Dawn: The Dawn of Agentic Superintelligence

Paper • 2609.15818 • Published 14 days ago • 429

Dream-RSI: Recursive Self-Improvement through Evolving Worlds

Paper • 2609.14858 • Published 14 days ago • 247

upvoted a paper 14 days ago

Language Self-Play For Data-Free Training

Paper • 2509.07414 • Published Sep 9, 2025 • 32

upvoted 2 papers 15 days ago

HyQuant: Hybrid-Precision Quantization for LLM Attention

Paper • 2608.27875 • Published Aug 28 • 30

Solar Open 2 Technical Report

Paper • 2607.20062 • Published Jul 24 • 6

upvoted a paper 16 days ago

Erase-then-Delta Attention: Decoupling Erase and Write Addresses in Delta-Rule Linear Attention

Paper • 2606.26560 • Published Jun 25 • 3