Track trials, findings, and clinical evidence as it lands.
Search papers, labs, and topics across Lattice.
This paper introduces the Prospective Intention Store (PIS), a novel approach that enhances prospective memory capabilities in small models by implementing typed intention stores for state tracking. By framing prospective memory as schema-constrained state tracking, the authors demonstrate that small models can outperform larger models on benchmark tasks without the need for complex training processes. The results show that DeepSeek-Chat with PIS achieves a Set-F1 score of 82.9% on PM-Bench, significantly surpassing previous benchmarks and showcasing the efficacy of the method in improving small model performance.
Small models can now outperform large models in prospective memory tasks, achieving a groundbreaking 82.9% Set-F1 score with a simple, training-free approach.
Prospective memory means carrying out a deferred intention at the right future cue while other work continues. Benchmarks now isolate it as an agent skill, yet frontier LLMs still struggle: the best published PM-Bench scaffold reaches only 65.1% Set-F1. We argue that this loop is schema-constrained state tracking rather than open-ended reasoning, and that small models can execute it when the action space is typed. We propose the Prospective Intention Store (PIS) that puts lifecycle logic in code and scoped language work on the model. The scaffold is agentic and training-free: no selector fine-tuning and no trajectory distillation. On PM-Bench, DeepSeek-Chat with PIS reaches 82.9% Set-F1. On Gemma-E2B, Set-F1 is only 4.2% without a store and at most 6.6% under seven retrospective memories, while PIS reaches 66.2%. PIS further reaches 70.1% Set-F1, where retrospective memory methods stay at most 54.4%. PIS sets a new state of the art on this benchmark and enables small models to surpass the published large-model scaffold.