Gage: Okay - no. You don't think I deserve your attention.
Mark Zuckerberg: I think if your clients want to sit on my shoulders and call themselves tall, they have the right to give it a try - but there's no requirement that I enjoy sitting here listening to people lie. You have part of my attention - you have the minimum amount. The rest of my attention is back at the offices of Facebook, where my colleagues and I are doing things that no one in this room, including and especially your clients, are intellectually or creatively capable of doing.
> As I write this, Bitcoin just past $16,000 in value. I would bet any amount of money that on December 7, 2018, the price of Bitcoin will be below that value.
You don't have to "would bet", you can short it on some exchanges and soon you'll be able to short using futures. Bears should put their money where their mouth is. Bulls sure are doing it.
Suggestion : let the user select a confidence level at which FaceID unlocks the phone. This would allow the user to decrease false positives at the cost of false negatives. Not very Apple to expose settings, but in this case it seems like a no Brainerd, much like choosing between 4 or 6 digit pins.
I tried making the switch from Chrome but ran into three issues :
1) Firefox does not import Chrome logins and passwords. Major pain.
2) Firefox freezes when importing Chrome history and bookmarks. I get a "script is taking too long" message. If I force close and reopen, then try to delete the partially imported content, it freezes as well with the same message. Tried deleting profile and restarting restarting to no avail.
3) (more of a nuisance than a real issue) I need to use the same browser on all devices so that everything is in sync. Google Search in Firefox Android looks like a page from the 90s and is missing Search Tools. Maybe Google isn't playing nice?
Did people migrating from Chrome experience issues 1) and 2)? Tried the initial import on Linux and macOS, same issue.
Having the results you are most interested in highlighted and the bad ones removed seriously upgrades the search experience. Would be nice if Google had this functionality built in. I know they had the remove results thing builtin back in the day, but I find highlighting trusted sites even more important.
https://github.com/kudkudak/word-embeddings-benchmarks has a pretty nice evaluation of existing embedding methods. Notably missing from this article is GloVe ( https://nlp.stanford.edu/projects/glove/) and LexVec ( https://github.com/alexandres/lexvec ) both which tend to outperform word2vec in both intrinsic and extrinsic tasks. Also of interest are methods which perform retrofitting, improving already trained embeddings. Morph fitting (ACL 2017) is a good example. Hashimoto et al (2016) sheds some interesting insight on how embeddings methods are performing metric recovery. Lots of exciting stuff in this area.
word2vec yields better representations than PMI-SVD. If you want a better explicit PMI matrix factorization, have a look at https://github.com/alexandres/lexvec and the original paper. Explains why SVD performs poorly.
If you are looking for word embedding for production use, checkout fasttext, lexvec, glove, or word2vec. Don't use the approach described in this article.
Mark Zuckerberg: I think if your clients want to sit on my shoulders and call themselves tall, they have the right to give it a try - but there's no requirement that I enjoy sitting here listening to people lie. You have part of my attention - you have the minimum amount. The rest of my attention is back at the offices of Facebook, where my colleagues and I are doing things that no one in this room, including and especially your clients, are intellectually or creatively capable of doing.