I’m a new Jetson owner but a long time tech nerd. After working through hello AI I’m wondering… how can you do training and inference on other types of data?
So far all I have seen is graphic data in movies and pictures which is fantastic don’t get me wrong… how would you go about using text or maybe even some electrical circuit input or some random numerical data you get the point how do you get different types of data into training and inference?
I haven’t figured out if this is a hard question or a dumb one yet… I’m leaning towards dumb but I’m not really sure…
Hi @charliebray , different domains of data (i.e. images/video, text, time-series) typically use different DNN architectures. Whereas a ResNet might be used to classify video, an NLP model like BERT or GPT can be used to classify text, or a time-series RNN/LSTM can be used on time-series data.
These use different training/inference scripts. The best way to go about this is to typically look on GitHub or Google for the task in your framework of choice, for example “time-series LSTM with PyTorch”:
Thank you and if you’re the dusty from the hello AI you kicked ass man. Thanks for all of it.