madhu009 / deep-math-machine-learning.ai Goto Github PK

Hello,
I am a avid follower of @Sentdex and his work at pythonprogramming.net. This guy puts on a lot of effort creating his videos and content.

I saw your post Chapter 3.1 : SVM from Scratch in Python. where you have explained svm from scratch along with this code. I see a uncanny resemblance between his code and yours.

I request you to either cite his website or come up with your own version of "SVM from scratch".
Thank you.

Hi,
I have a doubt about one part (In 20). After training the model we are expecting line of code
trained_embeddings = embeddings.eval()
Why are we expecting that randomly initiated array will get updated when we are not even using it as an input.
please point out where I am wrong and correct it, as far as I know following code

loss = tf.reduce_mean(tf.nn.nce_loss(nce_weights, nce_biases, Y, embed, num_sampled, voc_size))
# Use the adam optimizer
optimizer = tf.train.AdamOptimizer(1e-1).minimize(loss)

it will update the nce_weights with each iteration not embedding matrix and why we would think it will get updated since we are not even using it at any input.

Thanks
Gagan

Thank you firstly for the tutorial
I wanted to ask if it is possible to use the final embeddings to test out a word and return top 10 similar terms.

e.g

Top 10 Similar words given an input word

word="external"
word_vec = final_embeddings[dictionary[word]]
sim = np.dot(word_vec,-final_embeddings.T).argsort()[0:8]
for idx in range(8):
print (reverse_dictionary[sim[idx]])

I tried the page at Meetups_Notebooks/CNN_End2End/CatsvsDogs.ipynb, but got an error while creating the function "convert_to_tflite". The error is thrown on this line...

converter = tf.contrib.lite.TocoConverter.from_frozen_graph(path+'frozen_catsVSdogs45.pb', ['Input'], ['Softmax']])

May be you have a extra "]" at the end after "softmax".