DAWN-ICL: Strategic Planning of Problem-solving Trajectories for Zero-Shot In-Context Learning

TL;DR

DAWN-ICL introduces a strategic planning approach using Monte Carlo Tree Search to improve zero-shot in-context learning by effectively selecting problem-solving trajectories, outperforming traditional methods especially in diverse task scenarios.

Contribution

It reformulates ZS-ICL as a planning problem and proposes a demonstration-aware MCTS method with a novel Q-value function for better trajectory planning.

Findings

Outperforms existing ZS-ICL methods in various scenarios.

Effective in both in-domain and cross-domain tasks.

Achieves better results than human-annotated demonstrations in some cases.

Abstract

Zero-shot in-context learning (ZS-ICL) aims to conduct in-context learning (ICL) without using human-annotated demonstrations. Most ZS-ICL methods use large language models (LLMs) to generate (input, label) pairs as pseudo-demonstrations and leverage historical pseudo-demonstrations to help solve the current problem. They assume that problems are from the same task and traverse them in a random order. However, in real-world scenarios, problems usually come from diverse tasks, and only a few belong to the same task. The random traversing order may generate unreliable pseudo-demonstrations and lead to error accumulation. To address this problem, we reformulate ZS-ICL as a planning problem and propose a Demonstration-aware Monte Carlo Tree Search (MCTS) approach (DAWN-ICL), which leverages MCTS to strategically plan the problem-solving trajectories for ZS-ICL. In addition, to achieve effective and efficient Q value estimation, we propose a novel demonstration-aware Q-value function and use it to enhance the selection phase and accelerate the expansion and simulation phases in MCTS. Extensive experiments demonstrate the effectiveness and efficiency of DAWN-ICL on in-domain and cross-domain scenarios, and it even outperforms ICL using human-annotated labels. The code is available at https://github.com/RUCAIBox/MCTS4ZSICL.