Logging to loggings/classifier/nd-dryrun
creating model and diffusion...
creating data loader...
creating optimizer...
training classifier model...
----------------------------
| grad_norm     | 321      |
| param_norm    | 165      |
| samples       | 256      |
| step          | 0        |
| train_loss    | 7.34     |
| train_loss_q0 | 5.78     |
| train_loss_q1 | 7.62     |
| train_loss_q2 | 9.35     |
| train_loss_q3 | 6.79     |
| val_loss      | 4.1      |
| val_loss_q0   | 4.1      |
| val_loss_q1   | 4.52     |
| val_loss_q2   | 3.82     |
| val_loss_q3   | 3.93     |
----------------------------
----------------------------
| grad_norm     | 132      |
| param_norm    | 165      |
| samples       | 2.82e+03 |
| step          | 10       |
| train_loss    | 2.29     |
| train_loss_q0 | 2.29     |
| train_loss_q1 | 2.28     |
| train_loss_q2 | 2.41     |
| train_loss_q3 | 2.18     |
| val_loss      | 1.91     |
| val_loss_q0   | 1.65     |
| val_loss_q1   | 1.95     |
| val_loss_q2   | 2.48     |
| val_loss_q3   | 1.64     |
----------------------------
----------------------------
| grad_norm     | 6.6      |
| param_norm    | 165      |
| samples       | 5.38e+03 |
| step          | 20       |
| train_loss    | 1.69     |
| train_loss_q0 | 1.71     |
| train_loss_q1 | 1.6      |
| train_loss_q2 | 1.81     |
| train_loss_q3 | 1.63     |
| val_loss      | 1.8      |
| val_loss_q0   | 1.21     |
| val_loss_q1   | 2.01     |
| val_loss_q2   | 1.37     |
| val_loss_q3   | 2.77     |
----------------------------