AI Paper Atlas

Research map / Large language models / Instruction and fine-tuning

Cappy: Outperforming and Boosting Large Multi-Task LMs with a Small Scorer

Bowen Tan, Yun Zhu, Lijuan Liu, Eric Xing, Zhiting Hu, Jindong Chen

NeurIPS 2023 · Large language models › Instruction and fine-tuning · approach: Calibration and knowledge in LLM tuning · 4 citations (OpenAlex)

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 changedWhereMeasuredEffect
replace self-scoring with Cappy-large selectionfrozen FLAN-T5-XXL with 17 samples · BIG-BenchRouge-L: 32.02 -> 36.93Improved
increase samples from 4 to 17 for self-scoringself-scoring on FLAN-T5-XXL · BIG-BenchRouge-L: 31.15 -> 32.02No clear effect
increase samples from 4 to 17 for CappyCappy-large on FLAN-T5-XXL · BIG-BenchRouge-L: 33.64 -> 36.93Improved
add Cappy-large to finetuned FLAN-T5-SmallCappy-large · BIG-BenchRouge-L: 33.95 -> 39.74Improved
add Cappy-large to finetuned FLAN-T5-LargeCappy-large · BIG-BenchRouge-L: 46.49 -> 48.98Improved
add Cappy-large on top of in-context learningFLAN-T5-Large ICL · BIG-BenchRouge-L: 27.00 -> 31.84Improved
add Cappy-large on top of prompt tuningFLAN-T5-Large prompt tuning · BIG-BenchRouge-L: 38.43 -> 42.71Improved
Cappy scoring versus FLAN-T5-XXL self-scoring on codenames, simple_arithmetic_json, sufficient_informationCappy · BIG-Bench per-taskRouge-LHurt

Closest papers by meaning

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.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy