Rethinking On-Policy Distillation of Large Language Models: Phenomenology, Mechanism, and Recipe
Paper • 2604.13016 • Published • 71
Alan
hiyata
AI & ML interests
ML applications in virus-host network prediction, synthetic genome generation, and genomic dataset design.
Recent Activity
upvoted a paper 2 days ago
Strips as Tokens: Artist Mesh Generation with Native UV SegmentationOrganizations
None yet
upvoted a paper 1 day ago
upvoted a paper 2 days ago
upvoted 7
papers 3 days ago
DataFlex: A Unified Framework for Data-Centric Dynamic Training of Large Language Models
Paper • 2603.26164 • Published • 352
Adam's Law: Textual Frequency Law on Large Language Models
Paper • 2604.02176 • Published • 473
EXAONE 4.5 Technical Report
Paper • 2604.08644 • Published • 59
RefineAnything: Multimodal Region-Specific Refinement for Perfect Local Details
Paper • 2604.06870 • Published • 39
Matrix-Game 3.0: Real-Time and Streaming Interactive World Model with Long-Horizon Memory
Paper • 2604.08995 • Published • 44
FORGE:Fine-grained Multimodal Evaluation for Manufacturing Scenarios
Paper • 2604.07413 • Published • 91
WildDet3D: Scaling Promptable 3D Detection in the Wild
Paper • 2604.08626 • Published • 232
upvoted 2
papers 6 days ago
upvoted 3
papers 7 days ago
SEVerA: Verified Synthesis of Self-Evolving Agents
Paper • 2603.25111 • Published • 31
INSPATIO-WORLD: A Real-Time 4D World Simulator via Spatiotemporal Autoregressive Modeling
Paper • 2604.07209 • Published • 35
FP4 Explore, BF16 Train: Diffusion Reinforcement Learning via Efficient Rollout Scaling
Paper • 2604.06916 • Published • 34
upvoted 3
papers 8 days ago
Learning to Retrieve from Agent Trajectories
Paper • 2604.04949 • Published • 69
Claw-Eval: Toward Trustworthy Evaluation of Autonomous Agents
Paper • 2604.06132 • Published • 114
Video-MME-v2: Towards the Next Stage in Benchmarks for Comprehensive Video Understanding
Paper • 2604.05015 • Published • 232
upvoted 3
papers 9 days ago
MinerU2.5-Pro: Pushing the Limits of Data-Centric Document Parsing at Scale
Paper • 2604.04771 • Published • 119
FileGram: Grounding Agent Personalization in File-System Behavioral Traces
Paper • 2604.04901 • Published • 40
TriAttention: Efficient Long Reasoning with Trigonometric KV Compression
Paper • 2604.04921 • Published • 107