Model
UNet_Medical TensorFlow checkpoint trained with AMP
Use the NGC CLI to download:
Copied!
1 Version
20.06.0Selected08/21/2020 3:24 AM UTC506.43 MBAccuracy: 00 EpochsBatch Size: 8GPU: V100 Copied!
20.06.0Selected
08/21/2020 3:24 AM UTC506.43 MBAccuracy: 00 EpochsBatch Size: 8GPU: V100
Copied!
architecture
| Key | Value |
|---|---|
| task | pretraining |
| type | base |
performance
| Key | Value |
|---|---|
| eval_ce_loss | 0.2149832844734192 |
| eval_dice_loss | 0.1061079204082489 |
| eval_dice_score | 0.8938920497894287 |
| eval_total_loss | 0.3210912048816681 |
training
| Key | Value |
|---|---|
| batch_size | 8 |
| dataset | EM segmentation challenge |
| iterations | 6400 |
| learning_rate | 0.0001 |
| training_precision | AMP |