name: data-centric-env
description: >
  Train an LLM agent to improve ML datasets using the Data-Centric AI paradigm.
  The agent coordinates specialist sub-agents (CleanerAgent, AugmenterAgent,
  BalancerAgent, ValidatorAgent) to fix data quality issues and improve a fixed
  classifier's accuracy — without changing the model architecture or hyperparameters.
  Based on Andrew Ng's Data-Centric AI framework.
version: "1.1.0"
tags:
  - data-centric-ai
  - world-modeling
  - professional-tasks
  - reinforcement-learning
  - grpo
  - unsloth
  - curriculum-learning

tasks:
  - name: task_0_tutorial
    description: >
      Single-issue tutorial. 100 rows, binary classification, 4 features.
      Only issue: 15-30% missing values. Baseline ~0.62, target 0.73. Budget: 30 steps.
      Note: query_analyst costs 2 budget total (1 cmd step + 1 internal); query_validator also costs 2 total.
  - name: task_1_easy
    description: >
      Two issues: missing values (8-25%) + mild class imbalance (5-20%).
      200 rows, binary, 5 features. Baseline ~0.63, target 0.79. Budget: 25 steps.
  - name: task_2_medium
    description: >
      Four issues: missing values, duplicates, class imbalance, type errors.
      500 rows, 3-class, 7 features. Baseline ~0.58, target 0.74. Budget: 40 steps.
  - name: task_3_hard
    description: >
      Six issues: missing values, duplicates, imbalance, type errors, outliers,
      cross-column logic errors. 900 rows, 4-class, 10 features.
      Baseline ~0.54, target 0.71. Budget: 60 steps.

action_type: text
observation_type: structured

# Reward components (v2 — strict discrimination design):
#   accuracy_reward:    [-1.0, +0.80] — Δacc×2.5 mid-ep; submit: base+efficiency×budget_fraction+stretch
#   process_reward:     [-0.20, +0.13] — strict workflow enforcement (blind apply=-0.08, etc.)
#   preservation_reward:[-0.50, +0.05] — row preservation ≥92% required for bonus
#   efficiency_reward:  [-0.10, +0.25] — at submit only; accuracy_gain/budget_used × 3.0
#   step_reward:        [-0.30, +0.15] — proxy quality delta per apply (no classifier)
# Total clamped to [-1.0, 1.0] by compute_total_reward()
reward_range: [-1.0, 1.0]