Hi mentors,
in course 5 week 2 lecture, the correction for GloVe slide 2 is to make j=t and i=c.
If so, in slide 4 (and quiz q9), in the minimize formula, should theta_i_T e_j be theta_j_T e_i? Thank you so much for your response!
Hello, YFChiu.
Please share the time where you felt the need of replacing those values. I am not able to locate the errors that you have pointed out in the slides.
Hi @Rashmi, I actually do notice similar to this and I came in search for some reason.
In week 2, for the GloVe lecture. At the beginning of the lecture and even at the clarification before the lecture, Xij was said to be the number of times that j occurred in the context of i. (2:19)
To simplify this, using previous models like skip-gram, i is basically c(content) and j is t(target).
However, at time step (3:05), theta is expressed in form of i, and later i was labeled as the target, and not content as earlier described. This also affected j, and was treated as the content, instead of using it as a target.
I want to know if this is an error or just due to the fact that Xij is symmetric to Xji.
Thanks.
Hello Akingbeni David,
Thank you for your question.
Prof Ng has tried to describe the global vectors for word representation in the used text corpus consisting of 10,000 words.
For that, he starts with context (c) and target (t). He then says that Xij as being x subscript tc, which means, how many times does the word (i) appear in context of word (j). That again symbolizes that the number of times (t) appears in context of (c).
[i=t & j=c]
Confusion happens when Prof Ng changes the symbolic representations of Xij on the left-hand side as (c) & (t).
But here, we need to understand that he also says that ‘depending on the definition of context and target words’, one might have Xij equals to Xji and this is where (i) & (j) changes into (c) & (t).
So, there is no such error, but was a way of saying and represention.
Now, while minimizing the difference between theta i transpose e_j minus log X_ij squared through GloVe model, he has used the same representation again where i and j are playing the role of t and c.
I would suggest that you must go through the video again after having a nice coffee and you will definitely understand the representations better then 
1 Like
Somewhat annoying that this confusion translates directly into the quiz, where the “correct” answer (“word j appears in the context of word i”) is precisely the opposite of what is shown on the lecture slide.