{
 "setting": "KD + ground truth CE",
 "alpha": 0.5,
 "order": "math_code",
 "seed": 1,
 "exp": 2,
 "student": "Qwen/Qwen2.5-1.5B-Instruct",
 "gpu_student": "NVIDIA RTX A6000",
 "gpu_teachers": "NVIDIA RTX 6000 Ada Generation",
 "load_h": 0.003992888331413269,
 "stages": {
  "stage1_math": {
   "domain": "math",
   "steps": 896,
   "train_h": 3.425248695876863,
   "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": 896,
    "math_kl": 0.08436640837325561,
    "math_ce": 0.4846152951909727,
    "code_kl": 0.1533881911288058,
    "code_ce": 0.3149686746307666,
    "math_hard_kl": 0.09327549027114479,
    "math_hard_ce": 0.6137640068253667
   },
   "peak_gb_student": 36.57978296279907,
   "peak_gb_teachers": 30.618666172027588
  },
  "stage2_code": {
   "domain": "code",
   "steps": 922,
   "train_h": 3.420455579492781,
   "val_start": {
    "step": 0,
    "math_kl": 0.08435009805638696,
    "math_ce": 0.48463048857849117,
    "code_kl": 0.15332529827721086,
    "code_ce": 0.31500937218458064,
    "math_hard_kl": 0.09329687484267526,
    "math_hard_ce": 0.6136897417741203
   },
   "val_end": {
    "step": 922,
    "math_kl": 0.10094955866903507,
    "math_ce": 0.47546204311290524,
    "code_kl": 0.08989427999834172,
    "code_ce": 0.23964554178732264,
    "math_hard_kl": 0.12081789152162996,
    "math_hard_ce": 0.6168563930018784
   },
   "peak_gb_student": 33.86883306503296,
   "peak_gb_teachers": 30.005428314208984,
   "bench_h": 0.6256002970536549,
   "gpu_bench": "NVIDIA RTX A6000"
  }
 },
 "train_h": 6.845704275369644,
 "done": true
}