y0news
← Feed
Back to feed
🧠 AI NeutralImportance 4/10

Evaluating Zero-Shot and One-Shot Adaptation of Small Language Models in Leader-Follower Interaction

arXiv – CS AI|Rafael R. Baptista, Andr\'e de Lima Salgado, Ricardo V. Godoy, Marcelo Becker, Thiago Boaventura, Gustavo J. G. Lahr||7 views
🤖AI Summary

Researchers benchmarked small language models (SLMs) for leader-follower role classification in human-robot interaction, finding that fine-tuned Qwen2.5-0.5B achieves 86.66% accuracy with 22.2ms latency. The study demonstrates SLMs can effectively handle real-time role assignment for resource-constrained robots, though performance degrades with increased dialogue complexity.

Key Takeaways
  • Fine-tuned small language models achieve 86.66% accuracy in leader-follower role classification for human-robot interaction.
  • Zero-shot fine-tuning significantly outperforms baseline and prompt-engineered approaches for role assignment tasks.
  • Small language models maintain low latency (22.2ms per sample) suitable for real-time edge deployment.
  • Performance degrades in one-shot modes due to increased context length challenging model capacity.
  • SLMs offer viable alternative to large language models for resource-constrained mobile and assistive robots.
Read Original →via arXiv – CS AI
Act on this with AI
Stay ahead of the market.
Connect your wallet to an AI agent. It reads balances, proposes swaps and bridges across 15 chains — you keep full control of your keys.
Connect Wallet to AI →How it works
Related Articles