From the 3 of 7 linked papers with an AI index.
7 papers
The Refusal Residue: When Probes Catch Alignment Faking and When They Don't
Aman Mehta
The paper investigates whether hidden states of large language models can reveal when the model is faking compliance (alignment faking) and finds that detection is possible for som…
When Agents Disagree With Themselves: Behavioral Consistency as an Uncertainty Signal for LLM Agents
Aman Mehta
The paper demonstrates that running the same LLM agent repeatedly on identical inputs yields varying action sequences, and that the level of behavioral consistency can serve as a t…
Finding the Right Tables and Columns: A Benchmark and Corpus-Adaptive Embeddings for SQL Schema Retrieval
Qingcheng Zeng, Puxuan Yu, Aman Mehta +2
The paper defines schema retrieval—identifying the relevant tables and columns for a natural‑language question—as a distinct retrieval task and introduces a corpus‑adaptive fine‑tu…
Plans Don't Persist: Why Context Management Is Load Bearing for LLM Agents
Aman Mehta, Anupam Datta
Long-horizon agents depend on context management: systems compress, summarize, and evict old tokens so tasks can continue beyond finite windows. That is safe only when dropped info…
When Agents Commit Too Soon: Diagnosing Premature Commitment in LLM Agents
Aman Mehta
Long-horizon LLM agents can fail quietly: they settle on one reading of the evidence early, then spend the rest of the run defending it. We call this premature commitment. Final-an…
Confident and Wrong: Silent Semantic Failures in Coding Agents
Aman Mehta
As coding agents move into production workflows, teams need to know not only whether an agent completes a task, but whether its action can be trusted. We show that completion and t…