rnow-train-jsonl

What problem does it solve?

This Skill formats and validates train.jsonl data for ReinforceNow, ensuring each line is a properly structured JSON object ready for RL or SFT workflows.

Core Features & Use Cases

• Format: standardizes each line as a JSON object with fields like messages, rewards, metadata, variables, tools, docker, and docker_env.
• Validation: enforces required fields and cross-checks rewards/tools names against rewards.py and tools.py.
• Use Case: prepare large train.jsonl datasets for agent training and evaluation, including sandbox/docker entries when needed.

Quick Start

Prepare a train.jsonl file with one JSON object per line. Example format: {"messages": [{"role": "user", "content": "Hello"}], "rewards": ["accuracy"]}