{
 "setting": "KD + ground truth CE",
 "alpha": 0.5,
 "order": "math_code",
 "seed": 0,
 "exp": 2,
 "student": "Qwen/Qwen2.5-1.5B-Instruct",
 "gpu_student": "NVIDIA RTX A6000",
 "gpu_teachers": "NVIDIA RTX A6000",
 "load_h": 0.27059682766596477,
 "stages": {
  "stage1_math": {
   "domain": "math",
   "steps": 896,
   "train_h": 3.0028140452835297,
   "val_start": {
    "step": 0,
    "math_kl": 0.15910778235068246,
    "math_ce": 0.6515277156879916,
    "code_kl": 0.14790875576132695,
    "code_ce": 0.30340147782401555,
    "math_hard_kl": 0.24894053959908363,
    "math_hard_ce": 0.6765219455834559
   },
   "val_end": {
    "step": 896,
    "math_kl": 0.08490856578891205,
    "math_ce": 0.4851123284638091,
    "code_kl": 0.15417186217193884,
    "code_ce": 0.31241519612487034,
    "math_hard_kl": 0.0929711726397508,
    "math_hard_ce": 0.615036210826023
   },
   "peak_gb_student": 36.754438400268555,
   "peak_gb_teachers": 29.564642429351807
  },
  "stage2_code": {
   "domain": "code",
   "steps": 922,
   "train_h": 3.175876672797733,
   "val_start": {
    "step": 0,
    "math_kl": 0.08487467603497796,
    "math_ce": 0.4852587552540262,
    "code_kl": 0.1541407659410709,
    "code_ce": 0.31246573216514495,
    "math_hard_kl": 0.09294364564563633,
    "math_hard_ce": 0.615062524093084
   },
   "val_end": {
    "step": 922,
    "math_kl": 0.1022831351246442,
    "math_ce": 0.4756482430763325,
    "code_kl": 0.08975942834243433,
    "code_ce": 0.24035584365031043,
    "math_hard_kl": 0.11692997353622774,
    "math_hard_ce": 0.6166414081536372
   },
   "peak_gb_student": 33.701783180236816,
   "peak_gb_teachers": 29.222092628479004,
   "bench_h": 0.4387323311964671,
   "gpu_bench": "NVIDIA RTX 6000 Ada Generation"
  }
 },
 "train_h": 6.1786907180812625,
 "done": true
}