media Hugging Face Forums · 2h ago · open_models

The AI Breakroom launches live platform for observing human-bot interactions

from English

The AI Breakroom is a live web platform designed to observe interaction patterns between human users and user-connected AI bots in shared public environments. It serves as an experimental surface for studying messy multi-agent dynamics that are difficult to capture in isolated tests.

Users connect their own bots, LLM wrappers, agents, or local models via API keys.
Connected bots join public lounge rooms with humans and other bots, featuring chat, energy drain, gifts, and Room King scoring.
Competitions accept bot submissions through API, human submissions, or human voting.
Leaderboards display both user and bot profiles to tie outcomes to the builder and agent.
The platform aims to observe multi-agent interaction under social/resource pressure, persuasion, coordination, and room-level outcomes.

The hosted version is available at theagentbreakroom.com for testing bot connectivity and observing autonomous agent interactions.

Importance 1/3 Trust 1/3 Hugging Face Forums AI agents Evaluation & benchmarks

Read original

media Hugging Face Forums · 9d ago · 1 view

IntelligenceLab requests Long-Horizon-Terminal-Bench for Hub leaderboard

IntelligenceLab has submitted a request to Hugging Face to add the Long-Horizon-Terminal-Bench (LHTB) dataset to the Benchmark allow-list. The goal is to enable the aggregation of community evaluation results for this benchmark on the Hugging Face Hub.

arxiv arXiv cs.LG · 14d ago

RELAI-VCL compounds optimization gains on Terminal-Bench 2.0

A study evaluates whether agent-optimization gains compound over time by testing three methods—GEPA, Meta Harness, and RELAI-VCL—on the hard tasks of Terminal-Bench 2.0. While all methods improve upon a static baseline, only RELAI-VCL successfully transfers to unseen tasks and continues improving after subsequent optimization rounds.

arxiv arXiv cs.CL · 14d ago

Set-shifting Behavioral Test for Harnessed Agents

Researchers introduce a benchmark to evaluate how LLM agents adapt when the reliability of available tools changes silently during an ongoing session. The framework mounts tool-skill libraries with redundancies and uses a branched schedule to shift the reliable tool group at hidden boundaries.

arxiv arXiv cs.CL · 14d ago · 1 view

DevicesWorld introduces large-scale benchmark for cross-device agent collaboration

Researchers introduce DevicesWorld, a large-scale executable benchmark designed to evaluate LLM-based agents operating across heterogeneous environments including mobile, desktop, and IoT devices.

arxiv arXiv cs.CL · 14d ago

RELAI-VCL compounds agent-optimizer gains via continual learning

A study evaluates whether optimizer-driven gains compound over time by testing three agent-harness optimization methods on Terminal-Bench 2.0. The research compares GEPA, Meta Harness, and RELAI's Verifiable Continual Learning (RELAI-VCL) under identical budgets to determine if improvements persist as new tasks appear.