fsds_cleaning_env / configs /basic_rl.yaml
israaaML's picture
v2: curriculum scheduling, SFT pipeline, reward redesign, agent guide
16038fc
raw
history blame contribute delete
251 Bytes
# Basic RL training config (requires PyYAML: pip install pyyaml)
task_id: ecommerce_mobile
n_episodes: 50
agent: heuristic
base_url: http://localhost:8000
max_steps_per_episode: 18
log_dir: logs
log_interval: 10
seed: null
output_dir: training_output