AI Paper Atlas

Research map / Large language models / Agents, code and math

PASS: Probabilistic Agentic Supernet Sampling for Interpretable and Adaptive Chest X-Ray Reasoning

Yushi Feng, Junye Du, Yingying Hong, Qifan Wang, Lequan Yu

AAAI 2026 · Large language models › Agents, code and math · approach: LLM-driven embodied and driving agents

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add early-exit action (ablation)PASS · CAB-Eaccuracy: 88.60 -> 91.22Improved
add contrastive path-rank pretraining (ablation)PASS · CAB-Eaccuracy: 87.86 -> 91.22Improved
add expert knowledge warm-up (ablation)PASS · CAB-Eaccuracy: 88.89 -> 91.22Improved
probabilistic supernet sampling vs finetuned GPT-4o on same training dataGPT-4o (finetuned) · CAB-Eaccuracy: 81.82 -> 91.22Improved
PASS vs strongest baseline MedRAXMedRAX · CAB-Eaccuracy: 89.54 -> 91.22Improved
PASS vs MedRAX on safety-critical splitMedRAX · CAB-E safety-critical splitaccuracy: 89.79 -> 93.50Improved
PASS vs MedRAX hallucination rateMedRAX · CAB-E safety-critical splithallucination rate (%): 1.60 -> 1.60No clear effect
PASS vs LLaVA-Med on LLM-Judge scoreLLaVA-Med · CAB-ELLM-Judge score: 82.65 -> 84.28Improved
PASS vs MedRAX on SLAKE (zero-shot)MedRAX · SLAKEembedding similarity: 74.90 -> 79.55Improved
PASS vs CheXagent / GPT-4o finetuned on SLAKE similarityPASS · SLAKEembedding similarity: 81.82 (GPT-4o finetuned) vs 79.55 (PASS)Hurt
PASS has higher latency than single-pass modelsPASS · CAB-Elatency: LLaVA-Med 21.43 vs PASS 22.06Hurt
increase cost penalty lambda to 0.03 (ablation)PASS · CAB-Ecost / accuracy: cost down roughly 20%, accuracy around 88%Mixed

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy