AI Paper Atlas

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

AMS-IO-Bench and AMS-IO-Agent: Benchmarking and Structured Reasoning for Analog and Mixed-Signal Integrated Circuit Input/Output Design

Zhishuai Zhang, Xintian Li, Shilong Liu, Aodong Zhang, Lu Jie, Nan Sun

AAAI 2026 · Large language models › Agents, code and math · approach: Long-horizon LLM agents · 2 citations (OpenAlex)

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 structured intent graph + adaptor + knowledge base (full agent) vs direct LLMGPT-4o · AMS-IO-Bench (30 cases)DRC+LVS pass rate (%): 0 -> 63.33Improved
add the intent graph adaptor to KB + intent graph (ablation)AMS-IO-Agent (DeepSeek-V3) · AMS-IO-Bench (30 cases)DRC+LVS pass rate (%): 0 -> 76.67Improved
add structured intent graph to the knowledge base (vs direct SKILL / Python code generation) (ablation)AMS-IO-Agent (DeepSeek-V3) · AMS-IO-Bench (30 cases)DRC+LVS pass rate (%): 0 -> 76.67Improved
add knowledge base to intent graph and adaptor (ablation)AMS-IO-Agent (DeepSeek-V3) · AMS-IO-Bench (30 cases)DRC+LVS pass rate (%): 0 -> 76.67Improved
use Claude-3.7 or DeepSeek-V3 instead of GPT-4o as backboneAMS-IO-Agent · AMS-IO-Bench (30 cases)DRC+LVS pass rate (%): 63.33 -> 76.67Improved
increase benchmark difficulty from Simple to HardAMS-IO-Agent · AMS-IO-Bench difficulty levelsDRC+LVS passes (of 10): 10/10 -> 2/10 (GPT-4o), 4/10 (Claude-3.7), 3/10 (DeepSeek-V3)Hurt

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