Ido Levy
@dolev31
Research on LLM agents: what they pursue unasked, and how safely they act. IBM · Weizmann Institute of Science
2
Published Tools
21
Total Stars
0
Weekly Downloads
4
GitHub Followers
36
Public Repos
Published Tools
1 MCP Server1 Agentacross 2 categoriesdolev31
Swap one RAG pipeline stage, hold the rest fixed, and measure whether the gain survives composition. The framework behind "Retrieval Augmented Generation Compositionality": 450 configurations, eight benchmarks, corpora from 5K to 200K documents.
...moreMCP Serverai
211 dir
Ido Levy, Segev Shlomov, Avi Yaeli, Nir Mashkif, Asaf Adi, Yinon Goldstein, Eilam Shapira, Leshem Choshen
Distill an LLM agent's closed-set decisions into cheap tabular classifiers: TabSchema, TabSynth, TabHead.
Agentuncategorised
1 dir