Research map / Large language models / Agents, code and math
AutoSizer: Automatic Sizing of Analog and Mixed-Signal Circuits via Large Language Model (LLM) Agents
Xi Yu, Dmitrii Torbunov, Soumyajit Mandal, Yihui Ren
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 changed | Where | Measured | Effect |
|---|---|---|---|
| add circuit understanding (CU) module (ablation) | AutoSizer · AMS-SizingBench ablation (5 circuits) | FOM: 28.65 -> 35.47 | Improved |
| add self-reflection loop (SRL) (ablation) | AutoSizer · AMS-SizingBench ablation (5 circuits) | FOM: 25.32 -> 35.47 | Improved |
| add LLM-guided adaptive optimization engine instead of fixed BO (ablation) | AutoSizer · AMS-SizingBench ablation (5 circuits) | FOM: 27.00 -> 35.47 | Improved |
| add initial search space decision without circuit understanding (ablation) | AutoSizer (w/o CU) · AMS-SizingBench ablation (5 circuits) | FOM: 33.43 -> 28.65 | Hurt |
| use AutoSizer instead of LEDRO on Hard circuits | LLM sizing agents · AMS-SizingBench Hard | success rate: 80.0% -> 100% | Improved |
| use LLM-based agents instead of traditional optimizers | circuit sizing · AMS-SizingBench Hard | success rate: 50.0% (GA) -> 80.0% (LEDRO) | Improved |
| use LLM-based agents instead of traditional optimizers | circuit sizing · AMS-SizingBench Easy and Medium | FOM | No clear effect |
| add self-reflection to LLM search directions (ADO-LLM lacks it) | ADO-LLM · AMS-SizingBench Hard | success rate: 41.0% (ADO-LLM) -> 100% (AutoSizer) | Improved |
| use LLM-only sizing without explicit optimizer (EE-Sizer) | EE-Sizer · AMS-SizingBench Easy | runtime (s): 257.0 -> 756.2 | Hurt |
| allow expand/unfix action in outer loop on hard circuits | AutoSizer outer loop · AMS-SizingBench Hard | FOM change: +16.4% | Improved |
| narrow ranges action | AutoSizer outer loop · AMS-SizingBench Hard | FOM change: -3.8% | Hurt |
| use Fixed trust-region TuRBO | TuRBO · AMS-SizingBench Easy | success rate: 39.8% vs 93.3% (Genetic) | Hurt |
Closest papers by meaning
- AUTOCIRCUIT-RL: Reinforcement Learning-Driven LLM for Automated Circuit Topology Generation (ICML 2025 · similarity 0.88)
- FALCON: An ML Framework for Fully Automated Layout-Constrained Analog Circuit Design (NeurIPS 2025 · similarity 0.88)
- AnalogGenie: A Generative Engine for Automatic Discovery of Analog Circuit Topologies (ICLR 2025 · similarity 0.87)
- LaMAGIC: Language-Model-based Topology Generation for Analog Integrated Circuits (ICML 2024 · similarity 0.86)
- AnalogCoder: Analog Circuit Design via Training-Free Code Generation (AAAI 2025 · similarity 0.86)
- Retrieval-Guided Reinforcement Learning for Boolean Circuit Minimization (ICLR 2024 · similarity 0.85)
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.
