Towards Communication-Efficient Social Intelligence in Language Agents
TACT trains language agents to reach social goals with fewer tokens via teacher-assisted strategy and expression revisions.
Teacher-Assisted Communication Training (TACT) aims to raise social goal attainment for language agents while cutting communication cost. Strategy and expression specialists revise student actions; partner responses are sampled to choose a teacher reference that balances local goal support against action-token cost, then distilled on-policy. On SOTOPIA, TACT records the highest Goal among evaluated methods on All and Hard while using substantially fewer target tokens than SFT+SDPO. On AgentSense it improves goal success over the initial student and reduces target tokens and interaction messages.
- Specialists revise strategy and strip unnecessary expression detail.
- Partner-response sampling picks revisions by goal support versus token cost.
- SOTOPIA Goal is highest on All and Hard with fewer tokens than SFT+SDPO.
- AgentSense goal success rises while tokens and messages fall.
From the original article
1 min read at arxiv.orgSocially intelligent language agents must negotiate, coordinate, and resolve conflicting preferences while respecting the time and attention of both participants. Balancing these demands is challenging because agents must convey enough to address a partner's constraints and advance their goals without adding words that do not help the interaction.
By Linxiao Gong · arxiv.org
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.