OmniOPD: Logit-Free On-Policy Distillation via Speculative Verification
Paper • 2606.01476 • Published • 8
Zhuokai Zhao
zhuokai
AI & ML interests
Data-Efficient Learning, LLM Reasoning and Safety, Active Learning, Recommender System
Recent Activity
upvoted a paper 25 days ago
OmniOPD: Logit-Free On-Policy Distillation via Speculative Verification upvoted a paper about 2 months ago
DecodingTrust-Agent Platform (DTap): A Controllable and Interactive Red-Teaming Platform for AI Agents authored a paper 3 months ago
Synthetic Sandbox for Training Machine Learning Engineering AgentsOrganizations
upvoted a paper 25 days ago
upvoted a paper about 2 months ago
authored a
paper 3 months ago
upvoted a paper 3 months ago
authored 2
papers 6 months ago
upvoted a paper 6 months ago
authored a
paper 8 months ago
upvoted a paper 8 months ago
authored 10
papers 8 months ago
From Uncertainty to Trust: Enhancing Reliability in Vision-Language Models with Uncertainty-Guided Dropout Decoding
Paper • 2412.06474 • Published
Beyond Reward Hacking: Causal Rewards for Large Language Model Alignment
Paper • 2501.09620 • Published
Transfer between Modalities with MetaQueries
Paper • 2504.06256 • Published • 2
S'MoRE: Structural Mixture of Residual Experts for LLM Fine-tuning
Paper • 2504.06426 • Published • 2
CAFe: Unifying Representation and Generation with Contrastive-Autoregressive Finetuning
Paper • 2503.19900 • Published
Boosting LLM Reasoning via Spontaneous Self-Correction
Paper • 2506.06923 • Published
RecoWorld: Building Simulated Environments for Agentic Recommender Systems
Paper • 2509.10397 • Published • 8
StreamMem: Query-Agnostic KV Cache Memory for Streaming Video Understanding
Paper • 2508.15717 • Published • 1
Let it Calm: Exploratory Annealed Decoding for Verifiable Reinforcement Learning
Paper • 2510.05251 • Published • 8
Thought Communication in Multiagent Collaboration
Paper • 2510.20733 • Published • 15
upvoted a paper 8 months ago