Hey Guys,

I found some small typos in the assignment. I am listing them here, so that they can be rectified accordingly.

### Exercise 02 | Step 1

The loss function will be squared

~~Frobenoius~~Frobeniusnorm

### UNQ_C3 | compute_loss

loss i is the

~~sum_diff_squard~~sum_diff_squareddivide

### Exercise 03 | Step 2

Computing the gradient of loss

~~in~~withrespect to transform matrix R

### Exercise 4 | Calculate transformation matrix R

Using

~~those~~justthe training set, find the

### 3.3 Finding the most similar tweets with LSH

The number of planes. We use

~~log2(256)~~log2(625)to have ~16 vectors/bucket.

### Exercise 10

The id table at key 0 has 3

document indices

The first 5 document indices stored at key 0 of

id tableare [3276, 3281, 3282]

### Exercise 11

This is more of a query than a typo. In the `approximate_knn`

function, I don’t understand the use of `ids_to_consider_set`

. If instead of the following line of code `if new_id not in ids_to_consider_set:`

we use the following line of code `if new_id not in ids_to_consider_l:`

, then also, there will be no repeated ids. Also, in this case, we don’t have to maintain another data structure for storing the same thing, i.e., the ids. Am I missing something here?

Cheers,

Elemento