By Eigent AI
We build custom RL environments, training data and benchmarks for agentic and knowledge work, for research teams training agents.
Benchmarks for agentic and knowledge work.
AgenticRL-Bench tests whether agents can do the engineering work behind agentic RL, across the whole stack: training data and recipe selection, algorithm development, infrastructure debugging, monitoring and evaluation.
Open the postSamples from any category are available on request.
Ask for a sample30 Categories
Built by
(opens in a new tab)
Tasks, examples & goals
Reproducible task suites
Train, evaluate & iterate
Results become new seed data
Or email env-lab@eigent.ai
