Papers by Satvik Golechha
NICE: To Optimize In-Context Examples or Not? (2024.acl-long)
Copied to clipboard
| Challenge: | Recent work shows that in-context learning and optimization of in-const examples (ICE) can improve the accuracy of large language models on a wide range of tasks. |
| Approach: | They propose a task-specific metric called Normalized Invariability to Choice of Examples (NICE) metric measures the learnability of tasks from a given instruction and provides a heuristic to decide whether to optimize ICE for a new task. |
| Outcome: | The proposed metric predicts the utility of optimizing ICE for a given task compared to random ICE. |