Laurence, these videos are great! Thank you for making them! Some feedback: the text titles of the videos aren't easy to read and that makes it difficult to know the order to watch related videos. Consider changing the sequence of the items in the title from: "Title, Subject, Part" to "Subject, Part, Title". For example, this video would be:" Text Classification, Pt.2 | Designing a neural network".
Thanks for giving Part 1 a title beginning with P and Part 2 a title beginning with D. This means they will alphabetise themselves in a list about as far apart as possible, and maximise the amount of searching the human has to do "manually," by eye-ball. Thank goodness the machines are coming!
Heheh. Nice point. I should make playlists with them ordered to make it easier to figure out which comes after which. I'll speak with the social folks. :)
Had a question about this code. In cell 22 you separate out a subset of the original training data and call this validation set. And you fit only this data. Yet in cell 23 you evaluate the test_data which is already returned separately. So why the need for this extra validation set? If you train on the entire train_data and validate on the test_data, are not the two datasets already distinct?
Hi, I have a question, Is lemmatization and stemming applied to the dataset? If not, is there built-in tf/keras feature that can pre-process the data? Thanks.
I am not fully sure, but you can use the saveModel feature of TF to create files that contain your trained model. You can then download these files and run them on your computer. BTW this same youtube channel has more information regarding this feature.
I tried to do that this programming with my data from tweets, but I get error below, Someone can help me?? nvalidArgumentError: indices[0,39] = -1 is not in [0, 2000) [[Node: embedding_23/embedding_lookup = ResourceGather[Tindices=DT_INT32, _class=["loc:@training_12/Adam/gradients/embedding_23/embedding_lookup_grad/Reshape"], dtype=DT_FLOAT, validate_indices=true, _device="/job:localhost/replica:0/task:0/device:CPU:0"](embedding_23/embeddings, embedding_23/Cast)]]
Laurence, these videos are great! Thank you for making them! Some feedback: the text titles of the videos aren't easy to read and that makes it difficult to know the order to watch related videos. Consider changing the sequence of the items in the title from: "Title, Subject, Part" to "Subject, Part, Title". For example, this video would be:" Text Classification, Pt.2 | Designing a neural network".
This was very good! The embedding example was great!
THanks! :)
تحياتي الخالصة من الجزائر thank you very mutch
Thanks for giving Part 1 a title beginning with P and Part 2 a title beginning with D. This means they will alphabetise themselves in a list about as far apart as possible, and maximise the amount of searching the human has to do "manually," by eye-ball.
Thank goodness the machines are coming!
Heheh. Nice point. I should make playlists with them ordered to make it easier to figure out which comes after which. I'll speak with the social folks. :)
Thank you for sharing this informative video! 😺🖐
My pleasure! :)
Had a question about this code. In cell 22 you separate out a subset of the original training data and call this validation set. And you fit only this data. Yet in cell 23 you evaluate the test_data which is already returned separately. So why the need for this extra validation set? If you train on the entire train_data and validate on the test_data, are not the two datasets already distinct?
Thank u for this awesome video and series
Thanks, Abishek! :)
Hi, I have a question,
Is lemmatization and stemming applied to the dataset?
If not, is there built-in tf/keras feature that can pre-process the data?
Thanks.
Why 16 dimensions? Why not 3 or 160?
Hi! Have you found out that answer to this? I'd greatly appreciate any insight into this!
Thank you so much for the video! The link to the workbook is not working, could you please upload it again?
www.tensorflow.org/tutorials/keras/text_classification
AWESOME
Thanks!
Nice
Hi, how do I export the trained model from Codelab to my computer?
I am not fully sure, but you can use the saveModel feature of TF to create files that contain your trained model. You can then download these files and run them on your computer. BTW this same youtube channel has more information regarding this feature.
How can we use model.predict() in this case??
oh, no. The 14th video can't be show, could yo tell me why?
I tried to do that this programming with my data from tweets, but I get error below, Someone can help me??
nvalidArgumentError: indices[0,39] = -1 is not in [0, 2000)
[[Node: embedding_23/embedding_lookup = ResourceGather[Tindices=DT_INT32, _class=["loc:@training_12/Adam/gradients/embedding_23/embedding_lookup_grad/Reshape"], dtype=DT_FLOAT, validate_indices=true, _device="/job:localhost/replica:0/task:0/device:CPU:0"](embedding_23/embeddings, embedding_23/Cast)]]
Pokémon