loaded backbone from experiments/runs/p4_tiny_backbone_seed01_20261029232626/checkpoint missing: 35
/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 0.3192 lr 1.00e-05 gn 0.00 (5s)
step 50 loss 0.2609 lr 5.10e-04 gn 0.00 (37s)
step 100 loss 0.2838 lr 1.00e-03 gn 0.01 (60s)
step 150 loss 0.2172 lr 9.99e-04 gn 0.01 (82s)
step 200 loss 0.3022 lr 9.97e-04 gn 0.02 (103s)
step 250 loss 0.2631 lr 9.94e-04 gn 0.01 (125s)
step 300 loss 0.3198 lr 9.89e-04 gn 0.01 (147s)
step 350 loss 0.2695 lr 9.84e-04 gn 0.01 (169s)
step 400 loss 0.3633 lr 9.76e-04 gn 0.01 (191s)
step 450 loss 0.2708 lr 9.68e-04 gn 0.01 (212s)
  val loss by budget: {0: 0.4144, 1: 0.4143, 2: 0.4143, 4: 0.4146, 8: 0.4164}
step 500 loss 0.3310 lr 9.58e-04 gn 0.02 (238s)
step 550 loss 0.2803 lr 9.48e-04 gn 0.01 (260s)
step 600 loss 0.3321 lr 9.36e-04 gn 0.01 (282s)
step 650 loss 0.2449 lr 9.22e-04 gn 0.01 (304s)
step 700 loss 0.2686 lr 9.08e-04 gn 0.02 (326s)
step 750 loss 0.2743 lr 8.93e-04 gn 0.03 (348s)
step 800 loss 0.4066 lr 8.77e-04 gn 0.02 (370s)
step 850 loss 0.2539 lr 8.59e-04 gn 0.02 (392s)
step 900 loss 0.3607 lr 8.41e-04 gn 0.02 (420s)