😎 A Survey of Efficient Reasoning for Large Reasoning Models: Language, Multimodality, Agent, and Beyond
-
Updated
Jan 22, 2026
😎 A Survey of Efficient Reasoning for Large Reasoning Models: Language, Multimodality, Agent, and Beyond
Qwen3.8-27B TWIN-TURBO (NVFP4 GGUF, all-Q8_0 precision chain) tuned to its limits on RTX 5090 Laptop 24GB: 74-78 tok/s with built-in MTP + CUDA graphs, overthinking -93%, vision 3.9s, 160K stable context (cliff mapped). 10 rounds of measured tuning, rejected approaches documented, continuously updated.
AVR: Learning Adaptive Reasoning Paths for Efficient Visual Reasoning
[ICLR 2026] FROST: Filtering Reasoning Outliers with Attention for Efficient Reasoning
反内耗 Skill:把反复想、纠结、脑内循环拆解成具体问题,判断是否值得继续思考、现在能否行动、继续想能否产生新信息。
TED-style presentation about Overthinking built with PptxGenJS — by Mohamed Gharieb
A structured AI prompt for sorting thoughts, reducing overthinking, and separating facts, interpretations, fears, and hopes.
To associate your repository with the overthinking topic, visit your repo's landing page and select "manage topics."