Skip to main content
Sean's user avatar
Sean's user avatar
Sean's user avatar
Sean
  • Member for 5 years, 10 months
  • Last seen more than a month ago
295 votes
11 answers
233k views

What exactly are keys, queries, and values in attention mechanisms?

20 votes
6 answers
15k views

Why do we use masking for padding in the Transformer's encoder?

10 votes
2 answers
3k views

What is the difference between the Viterbi algorithm and beam search?

7 votes
1 answer
162 views

What exactly does the word "align" mean in the attention model?

7 votes
1 answer
2k views

How to convert a list of integers into a probability distribution such that the smaller the integer the larger the probability value?

7 votes
2 answers
2k views

Why are language modeling pre-training objectives considered unsupervised?

6 votes
1 answer
358 views

Why is it bad if the estimates vary greatly depending on whether we divide by N or (N - 1) in multivariate analysis?

5 votes
2 answers
2k views

Why don't we estimate the prior in a Naive Bayes' classifier?

5 votes
0 answers
179 views

Why is homoscedasticity (homogeneity of variance) important in neural network layers?

5 votes
1 answer
986 views

Are text generation models generative or discriminative?

5 votes
0 answers
403 views

What is a "log-bilinear regression model?"

5 votes
2 answers
4k views

Is a kernel function basically just a mapping?

5 votes
2 answers
687 views

Does the concept of "confidence" apply to text generation tasks?

4 votes
1 answer
664 views

One-hot encoding for duplicate words

3 votes
2 answers
1k views

What would it mean to select features in a "greedy" fashion?

3 votes
1 answer
1k views

Why does a rotation transformation have 3 degrees of freedom?

2 votes
1 answer
1k views

What is the difference between "greedy selection" and "sampling according to a distribution?"

2 votes
1 answer
653 views

What does it mean to represent a word as an "indicator vector" in natural language processing?

2 votes
1 answer
860 views

For BERT-based models, do we absolutely have to include the [CLS] and [SEP] special tokens in the input data?

2 votes
1 answer
36 views

How does recall help precision overcome "length-related problems?"

2 votes
1 answer
3k views

How to count the number of independent parameters in a Bayesian network?

2 votes
0 answers
144 views

Decoder's autoregressive output keeps outputting the same token

2 votes
1 answer
90 views

Why is it a good idea to initialize the bias vector to have small values in feedforward neural networks?

2 votes
0 answers
4k views

Time Series w/ Linear Regression

2 votes
1 answer
111 views

How to know which statistical test to use?

2 votes
1 answer
64 views

What exactly is "fundamental data" and "technical data?"

2 votes
1 answer
237 views

Shannon entropy with regards to independent random variables

2 votes
1 answer
108 views

What exactly is Word2Vec in the context of CBoW and Skip-gram?

2 votes
1 answer
134 views

What does it mean to "directly learn a distribution?"

2 votes
2 answers
138 views

Model performance graph is oscillating?