You can add location information to your Tweets, such as your city or precise location, from the web and via third-party applications. You always have the option to delete your Tweet location history. Learn more
I updated the text classification colab to make it start from a set of raw text files on disk instead of a TFDS dataset object. Much more didactic https://colab.research.google.com/drive/1XcMJqKcTLjduIqjysRd1jGdRaKdetc3X … You can use `tf.keras.preprocessing.text_dataset_from_directory` (in tf-nightly) to generate text datasets.pic.twitter.com/bxNcbjbnz4
Note how you can do end-to-end text classification only with Keras objects: - Get datasets from files with `text_dataset_from_directory` - Tokenize and index with `TextVectorization` layer - Embed and classify the tokenized text with a Model Same for images. Keras is end-to-end
That’s a useful change. Far more representative of the position typical users find themselves in. Keras is phenomenal, by the way.
Seed 42 works much better
François, i'm still confused... How could you use tf.keras text features with official tf.text?
Write a layer that uses tf.text. Writing layers is easy
I have a question, why did you add a Dropout layer after the Embedding Layer ?
Regularizing the embeddings work well on unseen data, especially when you are training embedding on the fly and not using pre-trained ones.
Leet seed. Nice.
Your seed is elite
What would you do, if you're interested in which words are relevant for the classification of the model? Is there a way to make the relevant words visible like in the computer vision example where we see the activation heatmap?
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.
