Deeper is Not Always Better: Mitigating the Alignment Tax via Confident Layer Decoding
Paper • 2606.21906 • Published • 18
Sining Zhoubian
SiningZhou
AI & ML interests
None yet
Recent Activity
published a model 9 months ago
SiningZhou/Qwen3-8B-ReST-RL published a model 9 months ago
SiningZhou/Qwen3-8B-VMOrganizations
None yet
upvoted a paper 1 day ago
published 2
models 9 months ago
updated 2
models 9 months ago
upvoted a paper 10 months ago
authored 3
papers 10 months ago
ReST-MCTS*: LLM Self-Training via Process Reward Guided Tree Search
Paper • 2406.03816 • Published • 1
SciGLM: Training Scientific Language Models with Self-Reflective Instruction Annotation and Tuning
Paper • 2401.07950 • Published • 4
ReST-RL: Achieving Accurate Code Reasoning of LLMs with Optimized Self-Training and Decoding
Paper • 2508.19576 • Published • 2