Research map / Large language models / Agents, code and math
PCB-Bench: Benchmarking LLMs for Printed Circuit Board Placement and Routing
Jindong Li, Lianrong Chen, Bin Yang, Jiadong Zhu, Ying Wang, Yuzhe Ma, Menglin Yang
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 |
|---|---|---|---|
| domain-specific QLoRA fine-tuning of Qwen2.5-7B-Instruct | Qwen2.5-7B-Instruct · Task 1 Placement Macro-Level CQ | accuracy (%): 84.36 -> 90.50 | Improved |
| retrieval augmentation (RAG) for Qwen2.5-7B-Instruct | Qwen2.5-7B-Instruct · Task 1 Placement Macro-Level CQ | accuracy (%): 84.36 -> 88.26 | Improved |
| domain-specific QLoRA fine-tuning of Qwen2.5-7B-Instruct | Qwen2.5-7B-Instruct · Task 1 Placement Macro-Level QA | SBERT: 0.2236 -> 0.4687 | Improved |
| RAG for Qwen2.5-7B-Instruct | Qwen2.5-7B-Instruct · Task 1 Placement Macro-Level QA | SBERT: 0.2236 -> 0.5050 | Improved |
| domain-specific QLoRA fine-tuning of Qwen2.5-7B-Instruct | Qwen2.5-7B-Instruct · Task 1 Placement Macro-Level QA | BERTScore: 0.7302 -> 0.6203 | Hurt |
| RAG for Qwen2.5-7B-Instruct | Qwen2.5-7B-Instruct · Task 1 Placement Macro-Level QA | BERTScore: 0.7302 -> 0.6145 | Hurt |
| QLoRA fine-tuning of Qwen2.5-7B-Instruct | Qwen2.5-7B-Instruct · Task 1 Routing Micro-Level CQ | accuracy (%): 82.50 -> 82.50 | No clear effect |
| RAG for Qwen2.5-7B-Instruct | Qwen2.5-7B-Instruct · Task 1 Routing Macro-Level CQ | accuracy (%): 94.12 -> 92.43 | Hurt |
| RAG for Qwen2.5-7B-Instruct | Qwen2.5-7B-Instruct · Task 1 Routing Micro-Level CQ | accuracy (%): 82.50 -> 85.61 | Improved |
| larger frontier model (GPT-5 vs GPT-5-nano) | GPT-5 · Task 2 Basic Wiring Error Identification | accuracy (%): 72.72 -> 90.90 | Improved |
| larger model (Qwen3-VL-235B-thinking vs Qwen3-VL-8B) | Qwen3-VL · Task 2 Basic Wiring Error Identification | accuracy (%): 45.45 -> 54.54 | Improved |
| larger model (Qwen3-VL-235B-thinking vs Qwen3-VL-8B) | Qwen3-VL · Task 3 design description | F1: 0.8675 -> 0.8405 | Hurt |
Closest papers by meaning
- PCBSchemaGen: Reward-Guided LLM Code Synthesis for Printed Circuit Boards (PCB) Schematic Design with Structured Verification (NeurIPS 2026 · similarity 0.89)
- CircuitSense: A Hierarchical Circuit System Benchmark Bridging Visual Comprehension and Symbolic Reasoning in Engineering Design Process (ICLR 2026 · similarity 0.88)
- EEE-Bench: A Comprehensive Multimodal Electrical And Electronics Engineering Benchmark (CVPR 2025 · similarity 0.86)
- PCEval: A Benchmark for Evaluating Physical Computing Capabilities of Large Language Models (NeurIPS 2026 · similarity 0.86)
- CircuitFusion: Multimodal Circuit Representation Learning for Agile Chip Design (ICLR 2025 · similarity 0.85)
- Functional Matching of Logic Subgraphs: Beyond Structural Isomorphism (NeurIPS 2025 · 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.
