In the lecture - https://www.coursera.org/learn/natural-language-processing-tensorflow/lecture/uskDE/notebook-for-lesson-3 - there appears to be a minor error at around the 0:55 - :57 (secs) mark. Instead of 26,709 words in the vocabulary, it should be (I think) 26,709 articles (number of sentences) in the database. The vocab length, in this case, is 28435 (including the 2 special tokens).
hi @Gagan_Kumar
the vocabulary has been recalled for the sentences only.
My apologies about being a pest…:-)… In the lecture, it is mentioned that ------ " …You can also see that there are 26,709 words in the vocab and the longest sequence has 39 tokens… "------
I’m just trying to point out that actually there are 28435 words in the vocab… The 26,709, I feel, refers to the total number of records(or entries in the sarcasm.json file)…