Research map / Large language models / Language, safety and interpretability
TEMPERA: Test-Time Prompt Editing via Reinforcement Learning
Tianjun Zhang, Xuezhi Wang, Denny Zhou, Dale Schuurmans, Joseph E. Gonzalez
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 test-time editing (query-dependent policy) vs query-agnostic policy (ablation) | TEMPERA · AG News | accuracy: +3.0 (84.3 vs 81.3) | Improved |
| add test-time editing (query-dependent policy) vs query-agnostic policy (ablation) | TEMPERA · SST-2 | accuracy: -0.1 (91.9 vs 92.0) | No clear effect |
| add instruction and verbalizer edits to exemplar-only editing (ablation) | TEMPERA · AG News | accuracy: +3.3 (85.5 vs 82.2) | Improved |
| add verbalizer edits (on top of instruction edits) (ablation) | TEMPERA · AG News | accuracy: +1.2 (85.5 vs 84.3) | Improved |
| add instruction edits (on top of exemplar and verbalizer edits) (ablation) | TEMPERA · SST-2 / MR | accuracy: +0.5 SST-2 (92.4 vs 91.9); +0.2 MR (88.4 vs 88.2) | Improved |
| increase in-context exemplars from 4 to 8 (ablation) | TEMPERA · AG News | accuracy: -2.1 (82.2 vs 84.3) | Hurt |
| increase in-context exemplars from 2 to 8 (ablation) | TEMPERA · SST-2 / MR | accuracy: SST-2 +0.8 (92.4 vs 91.6); MR +0.5 (88.4 vs 87.9) | Improved |
| increase example pool size from 8 to 32 (ablation) | TEMPERA · AG News | accuracy: +0.6 (84.7 vs 84.1) | Improved |
| TEMPERA vs RLPrompt (discrete prompt RL) on few-shot classification | TEMPERA · SST-2 / CR | accuracy: +1.8 SST-2; +3.9 CR | Improved |
| TEMPERA on additional NLI/paraphrase tasks (RTE, QNLI, SNLI, MNLI, MRPC) vs finetuning | TEMPERA · RTE / QNLI / SNLI / MNLI / MRPC | accuracy: RTE 60.3 vs 58.6; QNLI 57.4 vs 60.2; SNLI 56.4 vs 54.64; MNLI 45.2 vs 47.8; MRPC 74.0 vs 77.4 | Mixed |
| TEMPERA with 512 training examples per class vs finetuning | TEMPERA · AG News (512/class) | accuracy: 88.6 vs 89.5 | Hurt |
Closest papers by meaning
- Dialogue for Prompting: A Policy-Gradient-Based Discrete Prompt Generation for Few-Shot Learning (AAAI 2024 · similarity 0.88)
- TempSamp-R1: Effective Temporal Sampling with Reinforcement Fine-Tuning for Video LLMs (NeurIPS 2025 · similarity 0.86)
- Scaling Prompt Synthesis for Large Language Model Reasoning (ICML 2026 · similarity 0.86)
- Dynamic Prompt Optimizing for Text-to-Image Generation (CVPR 2024 · similarity 0.86)
- Evolving Prompts In-Context: An Open-ended, Self-replicating Perspective (ICML 2025 · similarity 0.86)
- Reward-Guided Prompt Evolving in Reinforcement Learning for LLMs (ICML 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.
