fsds_cleaning_env / configs /basic_rl.json
israaaML's picture
v2: curriculum scheduling, SFT pipeline, reward redesign, agent guide
16038fc
raw
history blame contribute delete
244 Bytes
{
"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"
}