{
 "setting": "ground truth SFT",
 "alpha": 0.0,
 "order": "code_math",
 "seed": 0,
 "exp": 2,
 "student": "Qwen/Qwen2.5-1.5B-Instruct",
 "gpu_student": "NVIDIA RTX 6000 Ada Generation",
 "gpu_teachers": "NVIDIA RTX A6000",
 "load_h": 0.0037869377930959064,
 "stages": {
  "stage1_code": {
   "domain": "code",
   "steps": 922,
   "train_h": 2.7816942036151886,
   "val_start": {
    "step": 0,
    "math_kl": 0.15902198004314483,
    "math_ce": 0.6553564874444496,
    "code_kl": 0.14797331712898912,
    "code_ce": 0.3033394576202258,
    "math_hard_kl": 0.24902600898986357,
    "math_hard_ce": 0.6769691187959327
   },
   "val_end": {
    "step": 922,
    "math_kl": 0.18373014144020175,
    "math_ce": 0.7970197911624136,
    "code_kl": 0.3449104032087439,
    "code_ce": 0.24659605407089333,
    "math_hard_kl": 0.2772132094080552,
    "math_hard_ce": 0.7910297951570997
   },
   "peak_gb_student": 33.80862855911255,
   "peak_gb_teachers": 30.005428314208984
  },
  "stage2_math": {
   "domain": "math",
   "steps": 896,
   "train_h": 2.688884120848444,
   "val_start": {
    "step": 0,
    "math_kl": 0.18375438443199724,
    "math_ce": 0.796609162952875,
    "code_kl": 0.34498801428341486,
    "code_ce": 0.24669437762251217,
    "math_hard_kl": 0.27730645571818935,
    "math_hard_ce": 0.7908673360534969
   },
   "val_end": {
    "step": 896,
    "math_kl": 0.2711174080266869,
    "math_ce": 0.48686714429648636,
    "code_kl": 0.23916414351690382,
    "code_ce": 0.26016022039706727,
    "math_hard_kl": 0.3349221330783336,
    "math_hard_ce": 0.6155070331563108
   },
   "peak_gb_student": 36.68444871902466,
   "peak_gb_teachers": 30.639182090759277,
   "bench_h": 0.4127069314320882,
   "gpu_bench": "NVIDIA RTX A6000"
  }
 },
 "train_h": 5.470578324463633,
 "done": true
}