Search papers, labs, and topics across Lattice.
This paper introduces Astar, a specialized model designed to automate the proposal of evolution directions for self-evolving industrial AI systems, addressing a critical bottleneck that traditionally relies on expert input. By leveraging historical iteration data, Astar overcomes challenges such as sparse supervision and noisy data through a robust data pipeline and a combination of supervised fine-tuning and reinforcement learning techniques. The model significantly outperforms human experts and general-purpose LLMs, achieving a success rate of 67.86% in real-execution evaluations and driving substantial improvements in key performance metrics for Alibaba's Lazada advertising system.
Astar not only surpasses human experts in proposing evolution directions but also automates the entire iteration process, leading to substantial revenue gains.
Modern AI systems advance through continuous iteration: a loop of proposing evolution directions, implementing code, training, and evaluation. While the latter three stages are increasingly automated, the starting point --- proposing effective evolution directions --- remains a critical bottleneck that still relies heavily on senior experts. In this work, we explore whether AI can take over this role. We find that general-purpose LLMs, even the advanced GPT-5.5, offer only generic and misaligned suggestions: the required expertise is accumulated through experience rather than explicitly codified, and thus hard to inject directly. To this end, we propose Astar, a training-based approach that learns a specialized evolution-guiding model from the abundant iteration histories of industrial systems. Realizing this idea, however, raises four challenges: sparse supervision, noisy data, a vast direction space, and prohibitively expensive verification. We address them along two fronts. On the data side, we design a pipeline that turns noisy historical commits into a large, clean evolutionary corpus via pairwise sample expansion and noise filtering. On the model side, we train the model through mid-training, SFT, and RL, guiding evolution direction generation with hierarchical hints and using the reward model in RL as a fast surrogate evaluator. Astar has been deployed in Alibaba's Lazada advertising system for evolution direction proposal. Astar-8B achieves a single-proposal success rate of 0.6786 in real-execution evaluation, far exceeding human experts (0.3229) and the strongest general-purpose LLM (0.3071). More importantly, Astar closes the loop and enables fully automatic iteration: it guided 20 consecutive iterations over two weeks, improving offline Hitrate@200 by 23.6%, while an online A/B test yielded relative lifts of 4.86% in GMV and 1.82% in advertising revenue.
