English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Teaching agents to follow complex written instructions has been an importantyet elusive goal. One technique for enhancing learning efficiency is languagereward shaping (LRS). Within a reinforcement learning (RL) framework, LRSinvolves training a reward function that rewards behaviours precisely alignedwith given language instructions. We argue that the apparent success of LRS isbrittle, and prior positive findings can be attributed to weak RL baselines.Specifically, we identified suboptimal LRS designs that reward partiallymatched trajectories, and we characterised a novel reward perturbation tocapture this issue using the concept of loosening task constraints. We providedtheoretical and empirical evidence that agents trained using LRS rewardsconverge more slowly compared to pure RL agents. Our work highlights thebrittleness of existing LRS methods, which has been overlooked in the previousstudies.

A Reminder of its Brittleness: Language Reward Shaping May Hinder Learning for Instruction Following Agents