loaded backbone from experiments/runs/p4_tiny_backbone_seed01_20261029232626/checkpoint missing: 35
readout_scale set to backbone residual rms = 5.310
/Users/simon-pierreboucher/Desktop/Recherche-UQO/ia/prisme-slm/training/train_editor.py:55: UserWarning: The given NumPy array is not writable, and PyTorch does not support non-writable tensors. This means writing to this tensor will result in undefined behavior. You may want to copy the array to protect its data or make it writable before converting it to a tensor. This type of warning will be suppressed for the rest of this program. (Triggered internally at /Users/runner/work/pytorch/pytorch/torch/csrc/utils/tensor_numpy.cpp:214.)
  return torch.from_numpy(acts).to(device), torch.from_numpy(np.ascontiguousarray(b)).to(device)
step 0 loss 0.3472 lr 1.00e-05 gn 1.32 (4s)
step 50 loss 0.1013 lr 5.10e-04 gn 0.40 (41s)
  val loss by budget: {8: 0.7307}
step 100 loss 0.0165 lr 1.00e-03 gn 0.16 (90s)
step 150 loss 0.0053 lr 9.78e-04 gn 0.02 (137s)
  val loss by budget: {8: 1.1005}
step 200 loss 0.0024 lr 9.14e-04 gn 0.01 (173s)
step 250 loss 0.0010 lr 8.15e-04 gn 0.00 (206s)
  val loss by budget: {8: 1.3595}
step 300 loss 0.0008 lr 6.89e-04 gn 0.00 (254s)
step 350 loss 0.0004 lr 5.50e-04 gn 0.00 (300s)
  val loss by budget: {8: 1.4991}
step 400 loss 0.0004 lr 4.11e-04 gn 0.00 (347s)
step 450 loss 0.0004 lr 2.85e-04 gn 0.00 (391s)
  val loss by budget: {8: 1.6038}
step 500 loss 0.0006 lr 1.86e-04 gn 0.00 (437s)
step 550 loss 0.0003 lr 1.22e-04 gn 0.00 (480s)
step 599 loss 0.0003 lr 1.00e-04 gn 0.00 (517s)
  val loss by budget: {8: 1.645}
run p4_overfit64_b8_seed01_20261029235111 done