TeamGym: A Gym for Team Decision-Making by AI Agents with Distributed Information
Paper • 2610.37571 • Published • 1
Swarm Intelligence
LLM working as a team
-
TokenGate: Adapting LLMs for Fast and Memory-Efficient Prefill with Learned Token Pruning
Paper • 2610.26754 • Published • 12Note Proxy loss design for token importance score, making discrete choice learnable as scalar for attention score bias.