_lpa_

LLM Kessler Syndrome.

Presumably the goal is to have a lot of chips at the end (more than you started with?). I would assume if you bet 10 chips, you get 10 for being right, and zero for not. So betting 0 chips all the time probably won't get you the job!

> And as usual, Benedict Evans has much better take on the situation. Mostly likely because of his major in History.

Which is?

> * Doing something illegal is wrong, but if someone aquired wealth legally, I don't see the inequality a problem.

If you are rich enough, you can influence what is illegal. I don't know of a system of government that is robust to this kind of problem.

Might be worth pointing out that fp16 on a k80 is woeful.

Can't you? I have just been using the lightning class as a wrapper, with my normal model on self.model or something like this. Same for the dataloaders.

I like the idea of kubernetes, and (some time ago) worked through a basic tutorial. My main confusion is how to set up a development environment. Are there any guides you could suggest that cover basic workflows?

I always had the impression that the incentive alignment between Apple and it's users was far better than any of the other big tech companies.

I'm curious - do you think affiliate links are a problem, and if so, why?

Yeah, my partner recently had to turn down a place at lambda for this reason. Hopefully you can sort the UK soon!

> histones are irrelevant, since prokaryotes don't have any

Good point :).

The keyword to search for, if you are interested, is epigenetics. Modification to e.g. histone protiens can change the expression of genes through a variety of mechanisms. As far as I'm aware, exactly how this works isn't that well understood.

This is a problem in most areas of scientific research. There is no incentive to publish negative results, so no one does.

I had a macro in emacs that wrapped the highlighted text in some xml tag (say <book></book>). Processing that I could label it however - e.g. IOB or whatever you fancy. The labelling didn't really take that long to do, maybe a few hours over a couple of days.

I did this a while ago (www.hnreads.com, bookbot.io). Manually labelled ~2k comments, trained a NER system, then validated the titles via amazons api. Was pretty easy once I had manually labelled comments. Don't recall my F1/precision/recall scores - they were ok but lower than the state of the art reported in papers.

We just had a paper accepted after 11 months. There is no good reason for it taking this long. I wish biology would adopt the publishing practices of physics or machine learning.

I've moved completely from Vim to Spacemacs. I can't see myself going back to plain Vim any time soon. The emulation via evil is very good, and the layers provided are generally immediately usable with consistent keybindings.

I haven't regretted doing my PhD in the slightest. I've learned a huge amount, in a wide variety of areas. However, the system in the UK is a bit more favourable regarding funding. Having said that, the funding doesn't scale with location. I live in one of the most expensive cities in the UK (outside London), and am funded to the same level as everyone else outside of London.

The bigger problem I have is in deciding to leave academia. There are still questions I would like to know the answer to (and I expect there always would be). It just doesn't seem to make sense to take that path. Low pay, high competition, and the likelihood of being on a series of 18month contracts for the majority of your career. Indeed, in my specific field (Bioinformatics) the career path isn't even that well defined, and you can more easily get a job at a software company who will pay several multiple of what you could expect as a postdoc. It just doesn't make sense to do it.

I like this idea! In my research we deal with relatively large amounts of sequence data, all of which needs to be associated with geo (https://www.ncbi.nlm.nih.gov/gds/). While geo is in many ways a good thing, it is not the most pleasant to use - I would love it if we could use something like torrents instead.

I feel like there is a danger, however, that using torrents would facilitate the thousands of nonstandard (often redundant) formats bioinformaticians seem to create.

I made something last winter (http://www.hnreads.com/), though I haven't fixed a couple of bugs or updated it since. I used named entity recognition rather than links to find book titles, perhaps something you might consider if you wanted to expand the site. I like the idea of showing the top books per time period!

I made http://www.hnreads.com a while ago to help find book suggestions from threads on hacker news. I expanded the idea to look at subreddits related to books, at http://www.bookbot.io - the eventual plan being to unify the two. There are quite a few things I would like to add, but they will have to wait until I finish my thesis!

In contrast to sentiment of many replies here, I have really enjoyed my PhD (admittedly I have not yet submitted - I plan to in the summer). I worked for a year prior to starting, and while that was fun, it felt very limited by comparison. Being around smart people who are interested in a variety of topics, and have the freedom to pursue them, makes for an amazing environment. I've learned a huge amount, and I think the tradeoff of low pay for a few years was easily worth it.

Asking for your money here refers both to a potential reader, and the the author of the article.

I'm coming to the end of my PhD, and have been contacted by a few recruiters for various hedge funds. It seems like a shame to leave academia. Realistically though, most people will be stuck in a series of ~18 month postdoc contracts, with little prospect of getting tenure. Add to this the poor pay, and hedge funds start to seem pretty attractive!

Well, the nature of NER is that it can extract 'Steve Jobs' as either a PERSON or a BOOK entity (in my case). You can in principal extract whatever kind of entity you want, provided you have appropriate training data. The idea (loosely) is that it learns the structure of text around, say, a book. You can then label a piece of text as a book, despite not previously knowing that title. It doesn't solve the problem of two books having identical titles, however :).

(Note: I'm not an expert in NER/NLP, so please correct me if I have got something wrong!)

I built a little tool to help with that, but pretty much the first result on amazon was always right - for books with more than one match, at least.

I might be a bit embarrassed with the code quality - this was one of my first attempts at front end development. I was planning to expand the features of the site a bit, including genre filters, so perhaps I could clean it up a bit and release it when I've done that.

Damn, sorry about that. Do you see a loading spinner on the page at all? I don't have an Android or BlackBerry, so wasn't able to test with either.

That's not really much to do with me - the credit there lies with fastly! This was the first time I've used a CDN, and I had a great experience using them. Highly recommended.

Yeah this is my mistake - sorry :(. I'll fix it as soon as I have time.