Research map / Large language models / Language, safety and interpretability
Advanced Black-Box Tuning of Large Language Models with Limited API Calls
Zhikang Xie, Weilin Wan, Peizhu Gong, Weizhong Zhang, Cheng Jin
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 |
|---|---|---|---|
| train proxy with GP-filter surrogate instead of offline Proxy-Tune | Qwen3-8B proxy / Qwen3-14B · 11-dataset average | accuracy (%): 84.03 -> 92.10 | Improved |
| GP-filter instead of CPT (about 1.58% of API calls) | Qwen3-8B proxy / Qwen3-14B · 11-dataset average | accuracy (%): 91.76 -> 92.10 | Improved |
| GP-random instead of CPT (6.94% of API calls) (ablation) | Qwen3-8B proxy / Qwen3-14B · 11-dataset average | accuracy (%): 91.76 -> 91.28 | No clear effect |
| diversity-filtered LogitMap Pairs instead of random sampling for GP training (ablation) | GP surrogate · 11-dataset average | accuracy (%): 91.28 -> 92.10 | Improved |
| GP-filter vs pretrained model | Llama2-13B · 11-dataset average | accuracy (%): 55.92 -> 86.85 | Improved |
| GP-filter vs LoRA-Tune | Llama2-13B · 11-dataset average | accuracy (%): 85.79 -> 86.85 | Improved |
| GP-filter vs CPT | Llama2-13B · 11-dataset average | accuracy (%): 86.41 -> 86.85 | Improved |
| GP-filter vs full fine-tune | Llama2-13B · 11-dataset average | accuracy (%): 88.58 -> 86.85 | Hurt |
| train GP with only 2 API calls (CoLA) (ablation) | GP-filter · CoLA | accuracy (%) | No clear effect |
| GP-filter vs pretrained | Mistral-7B, Qwen3, DeepSeek-R1-Distill · 11-dataset average | accuracy (%): +9.94, +7.24, +13.27 points | Improved |
Closest papers by meaning
- Enhancing Large Language Model Performance with Gradient-Based Parameter Selection (AAAI 2025 · similarity 0.90)
- BBox-Adapter: Lightweight Adapting for Black-Box Large Language Models (ICML 2024 · similarity 0.90)
- Boosting Multi-Domain Fine-Tuning of Large Language Models through Evolving Interactions between Samples (ICML 2025 · similarity 0.89)
- An Emulator for Fine-tuning Large Language Models using Small Language Models (ICLR 2024 · similarity 0.89)
- Data Mixing Optimization for Supervised Fine-Tuning of Large Language Models (ICML 2025 · similarity 0.89)
- SWIFT: A Scalable Lightweight Infrastructure for Fine-Tuning (AAAI 2025 · similarity 0.89)
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.
