metadata
license: mit
task_categories:
- other
tags:
- data-analysis
- agent
- reinforcement-learning
data_agent_harbor_train
5000 deterministic data-analysis tasks for agent RL (training split). Each task gives an agent a Kaggle dataset and a question; the answer is graded deterministically (exact -> numeric tolerance -> list/percent normalization -> symbolic, no LLM judge).
Difficulty tiers: {'easy': 3689, 'hard': 722, 'medium': 589}. Environments build from base image savatar101/env-data-agent-train:base.
Format
Harbor task suite: tasks/<id>/ (task.toml, instruction.md, environment/, tests/), registry.json, manifest.parquet.