Research map / Large language models / Language, safety and interpretability
Token-Efficient Change Detection in LLM APIs
Timothee Chauvin, Clément Lalanne, Erwan Le Merrer, Jean-Michel Loubes, Francois Taiani, Gilles Tredan
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 |
|---|---|---|---|
| use Border Inputs with only output tokens instead of black-box MET baseline | B3IT · TinyChange | average ROC AUC: 0.61 -> 0.9 | Improved |
| sample 5 prompts 10 times instead of one prompt 50 times (ablation) | B3IT · TinyChange | average ROC AUC | Improved |
| increase number of prompts at fixed cost (ablation) | B3IT · TinyChange | average ROC AUC | Improved |
| reduce fine-tuning to a single step (ablation) | B3IT · TinyChange fine-tuning scale | ROC AUC: 0.87 ROC AUC at 1 step | Mixed |
| add reasoning-budget query strategy for BI discovery (ablation) | B3IT BI search on OpenRouter endpoints · 131 endpoints | coverage with at least one BI (%): 76% -> 78% | Improved |
| search 2k prompts instead of 1k (ablation) | B3IT BI search · 131 endpoints | coverage with at least one BI (%): 73% -> 76% | Improved |
| raise sampling temperature above 0 for BI search (ablation) | B3IT BI search · 93-endpoint candidate set | BI coverage vs quality | Mixed |
| compare logprob-free BI test with grey-box LT | B3IT · TinyChange | average ROC AUC | No clear effect |
Closest papers by meaning
- TokenPowerBench: Benchmarking the Power Consumption of LLM Inference (AAAI 2026 · similarity 0.83)
- From Judgment to Interference: Early Stopping LLM Harmful Outputs via Streaming Content Monitoring (NeurIPS 2025 · similarity 0.82)
- ContractBench: Can LLM Agents Preserve Observation Contracts? (NeurIPS 2026 · similarity 0.82)
- Reducing sequential change detection to sequential estimation (ICML 2024 · similarity 0.82)
- Logits are All We Need to Adapt Closed Models (ICML 2025 · similarity 0.82)
- Optimal Online Change Detection via Random Fourier Features (NeurIPS 2025 · similarity 0.82)
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.
