outputs/train_v2/train.log

loading teacher pipeline (frozen)...

Loading pipeline components...:   0%|          | 0/5 [00:00<?, ?it/s]
Loading pipeline components...:  60%|██████    | 3/5 [00:00<00:00,  5.21it/s]

Loading weights:   0%|          | 0/398 [00:00<?, ?it/s]
Loading weights: 100%|██████████| 398/398 [00:00<00:00, 6296.85it/s]

Loading pipeline components...:  80%|████████  | 4/5 [00:00<00:00,  4.99it/s]
Loading pipeline components...: 100%|██████████| 5/5 [00:00<00:00,  6.03it/s]
loading v2 student (bf16; fp32-master is a noted upgrade)...
student: {'total_params': 3158317056, 'total_B': 3.158317056, 'n_double': 5, 'n_full_single': 14, 'n_surrogate': 6}
param groups: {'muon_tensors': 100, 'muon_params': 2963275776, 'adamw_tensors': 57, 'adamw_params': 195041280}
dataset: 200 samples; 300 steps, micro-batch 2 x accum 4 = eff 8

  0%|          | 0/4 [00:00<?, ?it/s]
 25%|██▌       | 1/4 [00:00<00:00,  7.11it/s]
 75%|███████▌  | 3/4 [00:00<00:00, 12.91it/s]
100%|██████████| 4/4 [00:00<00:00, 11.56it/s]
  step   20/300  loss 0.8337  0.32 it/s
  step   40/300  loss 0.8422  0.31 it/s
  step   60/300  loss 0.8529  0.31 it/s
  step   80/300  loss 0.8523  0.31 it/s
  step  100/300  loss 0.8263  0.31 it/s

  0%|          | 0/4 [00:00<?, ?it/s]
 75%|███████▌  | 3/4 [00:00<00:00, 15.60it/s]
100%|██████████| 4/4 [00:00<00:00, 13.86it/s]
  step  120/300  loss 0.8288  0.31 it/s
  step  140/300  loss 1.1654  0.31 it/s
  step  160/300  loss 2.4614  0.31 it/s
  step  180/300  loss 2.0306  0.31 it/s
  step  200/300  loss 1.6464  0.31 it/s

  0%|          | 0/4 [00:00<?, ?it/s]
 75%|███████▌  | 3/4 [00:00<00:00, 15.64it/s]
100%|██████████| 4/4 [00:00<00:00, 13.91it/s]
  step  220/300  loss 1.5967  0.31 it/s
  step  240/300  loss 1.6762  0.31 it/s
  step  260/300  loss 1.5040  0.31 it/s
  step  280/300  loss 1.4607  0.31 it/s
  step  300/300  loss 8.0790  0.31 it/s

  0%|          | 0/4 [00:00<?, ?it/s]
 75%|███████▌  | 3/4 [00:00<00:00, 15.71it/s]
100%|██████████| 4/4 [00:00<00:00, 13.98it/s]
saved trained student to outputs/train_v2/student_trained.pt; samples in outputs/train_v2/samples/
peak VRAM: 53.2GB
TRAIN_DONE rc=0