{
 "setting": "KD + ground truth CE",
 "alpha": 0.5,
 "order": "code_math",
 "seed": 0,
 "exp": 2,
 "student": "Qwen/Qwen2.5-1.5B-Instruct",
 "gpu_student": "NVIDIA RTX A6000",
 "gpu_teachers": "NVIDIA RTX 6000 Ada Generation",
 "load_h": 0.0037406995561387805,
 "stages": {
  "stage1_code": {
   "domain": "code",
   "steps": 922,
   "train_h": 3.215486346019639,
   "val_start": {
    "step": 0,
    "math_kl": 0.15874307828215575,
    "math_ce": 0.6515277156879916,
    "code_kl": 0.14786912634071003,
    "code_ce": 0.30340147782401555,
    "math_hard_kl": 0.24940142917923272,
    "math_hard_ce": 0.6765219455834559
   },
   "val_end": {
    "step": 922,
    "math_kl": 0.16229168284286724,
    "math_ce": 0.6282751871629569,
    "code_kl": 0.08942757638245312,
    "code_ce": 0.23860612096562683,
    "math_hard_kl": 0.25388608839478183,
    "math_hard_ce": 0.6804173316028584
   },
   "peak_gb_student": 33.80862855911255,
   "peak_gb_teachers": 30.005428314208984,
   "bench_h": 0.4878710929552714,
   "gpu_bench": "NVIDIA RTX A6000"
  },
  "stage2_math": {
   "domain": "math",
   "steps": 896,
   "train_h": 3.1273572798569997,
   "val_start": {
    "step": 0,
    "math_kl": 0.1623687996484868,
    "math_ce": 0.629426977682671,
    "code_kl": 0.08941266436979539,
    "code_ce": 0.23868094359472722,
    "math_hard_kl": 0.2538826426740015,
    "math_hard_ce": 0.6803296349017502
   },
   "val_end": {
    "step": 896,
    "math_kl": 0.08534254115593289,
    "math_ce": 0.4833468253707159,
    "code_kl": 0.10727858671355454,
    "code_ce": 0.2561915750177428,
    "math_hard_kl": 0.09355363536782725,
    "math_hard_ce": 0.6145162028758892
   },
   "peak_gb_student": 36.68444871902466,
   "peak_gb_teachers": 30.639182090759277,
   "bench_h": 0.6943173629045486,
   "gpu_bench": "NVIDIA RTX 6000 Ada Generation"
  }
 },
 "train_h": 6.342843625876639,
 "done": true
}