Accordion-Thinking: Self-Regulated Step Summaries for Efficient and Readable LLM Reasoning (https://arxiv.org/abs/2602.03249)
Zhicheng YANG
yangzhch6
AI & ML interests
reasoning with LLMs
Recent Activity
published a dataset 25 days ago
yangzhch6/mathnet-dsv4pro-proof upvoted a paper about 1 month ago
Attention Amnesia in Hybrid LLMs: When CoT Fine-Tuning Breaks Long-Range Recall, and How to Fix It updated a model about 1 month ago
yangzhch6/qwen3-4b-envfactory-nonthinkingOrganizations
None yet
models 45
yangzhch6/qwen3-4b-envfactory-nonthinking
4B • Updated • 2
yangzhch6/qwen3-4b-envfactory-thinking
4B • Updated • 1
yangzhch6/Qwen3-4B-Base-DeleThink
4B • Updated • 23
yangzhch6/Qwen2.5-Math-7B-DeleThink
8B • Updated • 6
yangzhch6/Qwen3-4B-Base-AccordionThinking-MixRL
4B • Updated • 151
yangzhch6/Qwen2.5-Math-7B-AccordionThinking-MixRL
8B • Updated • 2
yangzhch6/Qwen3-4B-n8-sglang-no_kl-grpo-0.5-1e-6-step340
4B • Updated • 6
yangzhch6/Qwen3-4B-n8-sglang-no_kl-grpo-0.5-1e-6-step300
4B • Updated • 2
yangzhch6/mcpfactory-qwen3-8b-newreward-step340
8B • Updated • 2
yangzhch6/mcpfactory-qwen3-8b-newreward-step300
8B • Updated • 2
datasets 17
yangzhch6/mathnet-dsv4pro-proof
Updated • 31
yangzhch6/Accordion-Thinking-Synthetic-Data
Viewer • Updated • 14.7k • 12
yangzhch6/DeepInformal-DeepTheorem-Synthetic
Viewer • Updated • 404k • 13 • 1
yangzhch6/DeepInformal-Openr1-Math-46K-Synthetic
Viewer • Updated • 165k • 20
yangzhch6/compare-openr1
Viewer • Updated • 45.8k • 10
yangzhch6/Align-Openr1-Math-46k
Viewer • Updated • 45.8k • 10
yangzhch6/DeepInformal-test
Viewer • Updated • 405 • 10
yangzhch6/DeepInformal-Putnam-1995-2024
Viewer • Updated • 356 • 11 • 1
yangzhch6/DeepInformal-DeepTheorem-DeepSeek-84k
Viewer • Updated • 84.1k • 26
yangzhch6/Putnam-Informal-1995-2024
Viewer • Updated • 360 • 12 • 1