redshellspy

[redshellspy @redshellspy]

@AnkaReuel @phillip_isola The link you have shared is for NeurIPS 2020 and not 2026.

[redshellspy @redshellspy]

@aryehazan even for positive weights, I prefer softmax 😶

[redshellspy @redshellspy]

@woshiyizhimao67 Ours is 3.33 😢. It also depends on AC as well, but I don't think it will go our way.

[redshellspy @redshellspy]

Just read NeurIPS automatic reference checker email. For the paper we submitted --- we got 0 "problematic references".

For an instant, I was like --- that's good but soon realized, paper that will be getting rejected, whats the point my emotion 🥲

[redshellspy @redshellspy]

@Anony34pm 4, 3, 3 after rebuttal.

[redshellspy @redshellspy]

@norpadon I read that paper and really liked it. One thing though -- if you represent your variational distribution using gaussian -- then I feel that traditional method where KL is computed exactly is more better. Note sure what is your use case though.

[redshellspy @redshellspy]

Just completed NeurIPS rebuttal.

Don't know why I'm doing this. May be I have some hope left😐

[redshellspy @redshellspy]

I thought of continuing the thread, but then realized I have to nothing to write.

This exists solely because ending at 3/n feels wrong.

4/4

[redshellspy @redshellspy]

I'm will be soon completing my masters and planning to apply for ML PhD programs in 2027 admission cycle (Hope luck is on my side)

Over the next year, I'm looking for a pre-PhD research position with a university professor or research group. 1/n

[redshellspy @redshellspy]

I reached out to professors and labs over past months. It didn't work out. Hoping that posting here may do some magic.

If your group has an opening, or if you know someone who might be looking for a pre-PhD researcher - I'd be grateful for any leads. 3/n

[redshellspy @redshellspy]

@GMAlexColovic exactly my thoughts when I saw the score today.

[redshellspy @redshellspy]

My ICML Submission experience : https://t.co/P2G21dGXfg

[redshellspy @redshellspy]

@bremen79 I feel that it makes much more sense to have Author AC comment deadline after final justification deadline. But here it's implemented in reverse.

[redshellspy @redshellspy]

@chanwoopark20 I found math exchange post in one of th comments. The result sqrt(k-1)/k is true for fronenius norm.

I don't know if we can get expression like this for other norms.