Ruochen Zhou

📍 Can LLMs discover, abstract, and reuse higher-level tool skills across tasks? Existing tool-use benchmarks test solving tasks with fixed tools. But real workflows contain recurring structures where efficiency comes from reusable tool compositions, not isolated calls. We introduce SkillCraft: 126 tasks across 6 domains designed to test whether LLM agents can acquire compositional skills, not just call atomic tools. We also propose Skill Mode, a lightweight protocol with four MCP primitives that let agents compose, verify, cache, and reuse tool chains at test time. Our Key findings across evaluating 8 SOTA models: ⚡Skill Mode enables agents to self-discover and reuse skills, leading to higher success and efficiency than agents without it. The gains are larger for stronger models. 🧠 Stronger models (e.g., Claude) discover more generalizable skills, which transfer across tasks and even across models. 🔍 Deeper composition ≠ better — shallow, well-tested skills generalize best. 🔗 Paper: arxiv.org/abs/2603.00718 💻 Code: github.com/shiqichen17/Sk… 🏠 Page: skillcraft-website.github.io/page (1/7)

🚀🔥 Thrilled to announce our ICML25 paper: "Why Is Spatial Reasoning Hard for VLMs? An Attention Mechanism Perspective on Focus Areas"! We dive into the core reasons behind spatial reasoning difficulties for Vision-Language Models from an attention mechanism view. 🌍🔍 Paper: arxiv.org/pdf/2503.01773 Code: github.com/shiqichen17/Ad… Website: shiqichen17.github.io/AdaptVis/

Knowledge Fusion of LLMs Is it possible to merge existing models into a more potent model? We have already seen a few ways that show the potential to effectively do this using approaches like weight merging and ensembling of models. This work proposes FuseLLM with the core idea of externalizing knowledge from multiple LLMs and transferring their capabilities to a target LLM. It leverages the generative distributions of source LLMs to externalize both their collective knowledge and individual strengths and transfer them to the target LLM through continual training. To put it simply, the idea is to benefit from the strengths of all the LLMs and combine them into one integrated model. Finds that the FuseLLM can improve the performance of the target model across a range of capabilities such as reasoning, common sense, and code generation. By the way, you can also perform the fusion among fine-tuned LLMs that specialize in specific tasks. This continues to be an interesting research area so hoping to document more on any new ideas and findings I come across.

Model editing has been an effective way to reduce hallucinations in LLMs, instead of undergoing resource-intensive retraining. 🤯However, our study, led by @JasonForJoy, @kaiwei_chang, & @VioletNPeng, reveals that current methods inadvertently impair the general skills of LLMs. 🔍Check for more details: Paper: arxiv.org/abs/2401.04700 Code: github.com/JasonForJoy/Mo… #LLM #LLMs #hallucination #GenerativeAI