agent-lightning-rl-training - Agent Lightning RL Training for Diverse Agents
Captures agent state snapshots and transitions, then decomposes episode returns for action-level reinforcement-learning updates across supported agent frameworks.
Tags
Updated: 2026-09-28Capabilities
Typical Inputs
Typical Outputs
What this skill does
- Capture agent state snapshots
- Report execution transitions
- Receive transition data
- Process RL training batches
- Decompose episode returns
- Assign action-level credit
- Train policies across frameworks
Inputs
- Agent framework
- Agent task
- State snapshots
- LLM actions
- Rewards
- Transition episodes
- Training model
Outputs
- Execution session ID
- Reported transitions
- Per-action rewards
- RL training updates