Model
Wide & Deep Base TensorFlow2 checkpoint trained with AMP on NVTabular preprocessed dataset
Use the NGC CLI to download:
Copied!
2 Versions
11/18/2021 10:11 PM UTC696.12 MBAccuracy: 020 EpochsBatch Size: 131072GPU: V100 Copied!
11/18/2021 10:11 PM UTC696.12 MBAccuracy: 020 EpochsBatch Size: 131072GPU: V100
Copied!
architecture
| Key | Value |
|---|---|
| task | pretraining |
| type | base |
performance
| Key | Value |
|---|---|
| MAP@12 | 0.6567 |
training
| Key | Value |
|---|---|
| batch_size | 131072 |
| dataset | Outbrain |
| epochs | 20 |
| lr_deep | 0.00012 |
| lr_wide | 0.02 |
| training_precision | AMP |
20.12.0Selected03/31/2021 12:37 AM UTC696.09 MBAccuracy: 020 EpochsBatch Size: 131072GPU: A100 Copied!
20.12.0Selected
03/31/2021 12:37 AM UTC696.09 MBAccuracy: 020 EpochsBatch Size: 131072GPU: A100
Copied!
architecture
| Key | Value |
|---|---|
| task | pretraining |
| type | base |
performance
| Key | Value |
|---|---|
| MAP@12 | 0.6567 |
training
| Key | Value |
|---|---|
| batch_size | 131072 |
| dataset | outbrain |
| epochs | 20 |
| lr_deep | 0.00012 |
| lr_wide | 0.02 |
| training_precision | AMP |