Research map / Large language models / Language, safety and interpretability
PrefixGPT: Prefix Adder Optimization by a Generative Pre-trained Transformer
Ruogu Ding, Xin Ning, Ulf Schlichtmann, Weikang Qian
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 pre-training on random valid adders (ablation) | PrefixGPT · 32-bit random-init optimization | final avg reward: 31.6% lower reward without pre-training | Improved |
| add RoPE to coordinate embedding (ablation) | PrefixGPT · 32-bit random-init optimization | final avg reward: 16.6% lower reward without RoPE | Improved |
| add KL regularization and best-design retrieval (ablation) | PrefixGPT · 32-bit random-init optimization | final avg reward: 1.3% decrease without | No clear effect |
| add KL regularization and best-design retrieval (ablation) | PrefixGPT · 32-bit random-init optimization | reward std dev at final iteration: 0.038 -> 0.018 | Improved |
| use PrefixGPT instead of ArithTree | adder optimizer · 48-bit mean ADP | mean ADP: 79.1% reduction | Improved |
| use PrefixGPT instead of ArithTree | adder optimizer · 48-bit ADP std | ADP std dev: 422.2 -> 25.2 | Improved |
| use PrefixGPT instead of ArithTree/PrefixRL | adder optimizer · prefix graph size, 12 bit-width/depth pairs | size: max 59.1% reduction (48-bit Ko.) | Improved |
Closest papers by meaning
- Scalable and Effective Arithmetic Tree Generation for Adder and Multiplier Designs (NeurIPS 2024 · similarity 0.85)
- Optimal Control for Transformer Architectures: Enhancing Generalization, Robustness and Efficiency (NeurIPS 2025 · similarity 0.81)
- CODA: Rewriting Transformer Blocks as GEMM-Epilogue Programs (NeurIPS 2026 · similarity 0.81)
- AnalogCoder: Analog Circuit Design via Training-Free Code Generation (AAAI 2025 · similarity 0.80)
- AnalogGenie-Lite: Enhancing Scalability and Precision in Circuit Topology Discovery through Lightweight Graph Modeling (ICML 2025 · similarity 0.80)
- GPU-Accelerated Synthesis of Mixed-Boolean Arithmetic: Beyond Caching (NeurIPS 2026 · similarity 0.80)
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.
