Training
Train a model with AgentEnv, using the training frameworks you are already familiar with
TRL: train with Hugging Face's TRL training framework, a model that plays your tasks or the agents they deploy.
Unsloth: train with Unsloth's GRPO on one GPU, through the TRL helper.
NeMo RL: train with NVIDIA's NeMo RL, through NeMo Gym.
prime-rl: train with Prime Intellect's prime-rl, through verifiers.
verl: train with verl, through AgentEnv's agent loop.
SkyRL: train with SkyRL, through AgentEnv's env and generator.
Miles: train with RadixArk's Miles, through AgentEnv's agent function.
slime: train with slime, through AgentEnv's generate function on its OpenAI adapter.
Last updated on




