Can we combine integer linear programming with exemplar selection to improve In-Context Learning?
Yes! All you need is to optimize your Knapsack π
The paper by
@TongletJ
et al. on
#SEER
was just accepted to
#EMNLP2023
β learn more in this 𧡠(1/8)
π°
It was observed that the performance of ICL depends heavily on the selection of the exemplars.
@TongletJ
et al. show how this combinatorial optimization problem can be formulated as a Knapsack Integer Linear program and optimized efficiently with deterministic solvers. (2/π§΅)
@TongletJ
In their
#EMNLP2023
paper the authors use a capacity constraint to control the size in tokens of the prompt and diversity constraints to favor the selection of exemplars β sharing the same reasoning properties as the test problem. (4/π§΅)
@TongletJ
They propose
#SEER
, a method to automatically generate a Knapsack program for
#HybridQA
problems. It achieves superior performance to exemplar selection baselines on the FinQA and TAT-QA datasets. (5/π§΅)
#EMNLP2023
@TongletJ
Tokens are the main unit price for commercial LLMs. Thanks to capacity constraints,
#SEER
directly optimizes the prompt size to meet restricted token budgets. (6/π§΅)
#EMNLP2023