Model
efficientnet-widese-b4 ImageNet pretrained weights
Use the NGC CLI to download:
Copied!
1 Version
20.12.0_ampSelected09/22/2022 6:02 PM UTC133.96 MBAccuracy: 0400 EpochsBatch Size: 64GPU: A100 Copied!
20.12.0_ampSelected
09/22/2022 6:02 PM UTC133.96 MBAccuracy: 0400 EpochsBatch Size: 64GPU: A100
Copied!
architecture
| Key | Value |
|---|---|
| type | efficientnet-widese-b4 |
performance
| Key | Value |
|---|---|
| val_top1 | 83.268 |
| val_top5 | 96.422 |
training
| Key | Value |
|---|---|
| data_backend | pytorch |
| model_config | fanin-sd0.8 |
| batch_size | 64 |
| data | /dev/shm/imagenet |
| num_classes | 1000 |
| image_size | 380 |
| run_epochs | 80 |
| optimizer_batch_size | 4096 |
| arch | efficientnet-widese-b4 |
| workers | 16 |
| epochs | 400 |
| early_stopping_patience | -1 |