{
  "sft_ckpt": "/ephemeral/ckpts/sft.pt",
  "prompt_path": "/ephemeral/data/rl_prompts_train.jsonl",
  "eval_prompt_path": "/ephemeral/data/rl_prompts_test.jsonl",
  "curriculum_path": "/ephemeral/data/arithmetic_prompts.jsonl",
  "curriculum_iters": 100,
  "out_ckpt": "/ephemeral/ckpts/grpo.pt",
  "iterations": 1000,
  "prompts_per_iter": 8,
  "group_size": 8,
  "rollout_len": 300,
  "temperature": 1.0,
  "top_p": 1.0,
  "grpo_epochs": 1,
  "clip": 0.2,
  "kl_coef": 0.04,
  "lr": 1e-06,
  "grad_clip": 1.0,
  "eval_every": 50,
  "save_every": 100
}