Just getting into Tensorflow and i want to run the tensorflow recommender tutorials with my own data.
However, the datastructures there (dict-like, tensors can be addressed by “keys”) do not fit with the datasets structures in the tf.data starters.
Can pls someone point me somewhere there i can get a better overview over the dataset types (sub-classes) and best to work with recommenders?
I am not sure exactly what you want.
You can get Tensor from csv file via pandas like this.
df = pd.read_csv(‘movie.csv’)
tf.data.Dataset.from_tensor_slices(dict(df))
Thanks, thats it already.
I was before only stumbling over examples there the df transformed into an array was a tensor itself, no input using the dict transformation on the df, tensors defined with a name that did not show up if looking at the instances etc.