NVIDIA - Digit
J'aime 0
Nom | NVIDIA - Digit |
---|---|
Version of the software | 6.0 |
Underlaying Framework | Tensorflow |
Development Activity | Important |
GPUs | CUDA compatible |
DL MODEL EDITOR | |
Graphical editor provided | Non Changing the model is possible by editing the network graph" |
DL models supported | |
- CNN | Oui |
- LSTM | |
- MLP / Auto-encoders | |
- RNN | |
Pretained Networks | Oui |
Use Keras API | Non |
Save Keras HDFS file (.h5) | |
Save Keras model .yaml file | |
Save Keras model file | |
Load Keras model .yaml file | |
Load Keras HDFS file (.h5) | |
Code generation | |
- Theano | |
- Tensorflow/Keras | |
- Caffe (v1) | |
Custom cost function | |
Load Keras Models + Weights | |
Load Caffe Models + Weights | |
Load Keras Code | ??? |
Model checker | Non |
Datasets Management | Oui |
Experiments Mangement | |
- Exp comparison | |
- Prev. Exp loader | |
TRAINING | |
DL model training | Oui |
Tune learning parameter | |
Live performance monitoring | |
INFERENCE | |
Inference available | Oui |
Results browser | |
View Valid/test set inference | |
Confusion matrix | |
Visualize intermediate layer activation | |
User test int | |
DEPLOIEMENT | |
- Rest API | |
PLATFORM | |
- Cloud | |
but no computing environnement" | |
- Cloud Trial period | Oui |
- Desktop | Oui |
OTHER | |
AutoML | Non |
Browser | |
Environment management | |
Python shell | |
Jupyter notebook | |
Documentation | |
(Keras API)" | |
(SONY Neural Network Library)" | |
Tutorials | |
Web Site | nvidia.com/... |
Avis utilisateurs et commentaires