Subcircuit Identification on AMS Hard (>40 transistors) 1.0
95.5F1 ScoreHeaRT
Evaluation Results
| Method | Links | |
|---|---|---|
| HeaRTLLM Backbone=Claude-4.6-Sonnet, Approach=HeaRT2025.11 | 95.5 | |
| HeaRTLLM Backbone=GPT-5, Approach=HeaRT2025.11 | 94.9 | |
| HeaRTLLM Backbone=DeepSeek-V3.2, Approach=HeaRT2025.11 | 94.6 | |
| HeaRTLLM Backbone=Gemini-2.5 Pro, Approach=HeaRT2025.11 | 91.4 | |
| Few-shot (Text)LLM Backbone=Claude-4.6-Sonnet, Approach=Few-shot (Text)2025.11 | 68.4 | |
| Few-shot (Image)LLM Backbone=Claude-4.6-Sonnet, Approach=Few-shot (Image)2025.11 | 65.7 | |
| Few-shot (Text)LLM Backbone=GPT-5, Approach=Few-shot (Text)2025.11 | 57.8 | |
| Few-shot (Image)LLM Backbone=GPT-5, Approach=Few-shot (Image)2025.11 | 54.1 | |
| Few-shot (Text)LLM Backbone=DeepSeek-V3.2, Approach=Few-shot (Text)2025.11 | 53.9 | |
| Few-shot (Text)LLM Backbone=Gemini-2.5 Pro, Approach=Few-shot (Text)2025.11 | 53.2 | |
| Few-shot (Image)LLM Backbone=DeepSeek-V3.2, Approach=Few-shot (Image)2025.11 | 49.3 | |
| Few-shot (Image)LLM Backbone=Gemini-2.5 Pro, Approach=Few-shot (Image)2025.11 | 46.9 | |
| HeaRTLLM Backbone=LLaMA-3.3-70B-Instruct†, Approach=HeaRT2025.11 | 33.6 | |
| Few-shot (Text)LLM Backbone=LLaMA-3.3-70B-Instruct†, Approach=Few-shot (Text)2025.11 | 16.7 |