nborwankar

8 ms·

31.
by nborwankar 3y ago
Is this “Apple silicon only” right now?
32.
by nborwankar 3y ago
<clap emoji> * 3
33.
by nborwankar 3y ago
Many people are going to complain about this post … to the manager ;-)
34.
by nborwankar 3y ago
Very interesting tgat you mention bad math teachers. I agree. During my teaching assistant stint during graduate studies in applied and pure math, I had one student who was failing. Talking to her I learned that a math teacher in high scho
35.
by nborwankar 3y ago
AI models compete with movie script writers. I believe the current writers strike in Hollywood includes ChatGPT4 related issues amongst many others.
36.
by nborwankar 3y ago
Thanks much for the offer. Will do in the next 2 weeks.
37.
by nborwankar 3y ago
To properly categorize these drugs we need to expand our thinking . Hmmm :-)
38.
by nborwankar 3y ago
Alternately, if they have the “right skills” why is this happening? Clearly they don’t have some of the needed skills to prevent this. And a self-DDOS might be a good indication of that.
39.
by nborwankar 3y ago
You may not, there are communities though that might. People who live in the “blast radius” of a polluting entity like a refinery might want to aggregate data over time to see if there’s a slow leak or whether the entity is emitting excessi
40.
Cadence uses Generative AI in chip design(seekingalpha.com)
3 points by nborwankar 3y ago|0 comments
41.
by nborwankar 3y ago
Cloud VM possible, not sure. Is it possible to run on HuggingFace without compromising data privacy?On-premise preferred. Capital cost to get to “Hello world” will probably convince principals to use HuggingFace if data privacy concerns can
42.
by nborwankar 3y ago
Thanks! Appreciate the detailed info on GPU’s and cost very much.
43.
by nborwankar 3y ago
Yes mostly answering questions and also summarizing and exploring by asking for explanations (understanding there could be non trivial amounts of hallucination) Really appreciate you offering advice. Will book time after the long weekend.
44.
by nborwankar 3y ago
The LLM needs to be run locally. So it can’t be a GPT API or an API for some cloud LLM. The docs can’t leave the perimeter. It can be a Pretrained model that is being “extended” (is there such a thing?) with the new docs. Alternately it can
45.
Ask HN: How to train and host an LLM for internal docs
8 points by nborwankar 3y ago|11 comments
46.
by nborwankar 3y ago
Importantly it’s a vector space with a metric. Not just an array of vectors. The original Word2Vec paper had the jaw dropping “king - man + woman -> queen” which needs the ability to do addition/subtraction on vectors. Also words “c
47.
by nborwankar 3y ago
I remember TinTin from high school in India in the 1970’s.
48.
by nborwankar 3y ago
Per capita emissions are a distraction. The earth’s atmosphere doesn’t care how many people were involved. What matters is total emissions. On that front China leads by a factor of 4 or 5 over India and the U.S. which are roughly the same.
49.
by nborwankar 3y ago
Nothing was ever free. Even Usenet. Universities and Internet savvy corporations ran Usenet servers. That’s why it was free to the rest of us. I tried running a Usenet node in 1995 or so on a 486/33 getting a feed from UC Berkeley. Had
50.
by nborwankar 3y ago
Meta question - why do we need the latest version of a text book every year. None of the undergraduate fundamental subjects change every year. It’s a racket - problems and some content are modified a bit and moved around making last year’s
51.
by nborwankar 3y ago
Just guessing but you have probably not traveled all over India where each state has a different cuisine and culture and language.
52.
by nborwankar 3y ago
Alternately put up links to local businesses interspersed with and if possible matched to chat topics. Count the click throughs and email a weekly report to businesses if the numbers are significant. You could also track how often the name
53.
by nborwankar 3y ago
His product vision is to make it into a one stop application with all media and e-commerce - something he calls X. The whole free speech platform thing is a distraction. There’s no money in free speech. It’s just a way to lure maximum numb
54.
by nborwankar 3y ago
This line of thinking may be confusing “sufficient” with “necessary”. I don’t believe Chomsky’s thesis and Kapatsinki’s statement are mutually exclusive. They could both be true. Chomsky didn’t appear to have made a general statement about
55.
by nborwankar 3y ago
The elephant in the room is enforcement. For example, how on earth are you going to control what a rogue state like North Korea might do with AGI supplied by say China for military use such as autonomous killer drones? Will the internationa
56.
by nborwankar 3y ago
Attention and the Transformer make it possible to recognize that the probability of “black” applying to the cat is much much higher than to the mat due to the phrasing “which is” in between the cat and black.
57.
by nborwankar 3y ago
There is annotated version of the paper. It also has a link to a GitHub repo in there. Google “The annotated transformer”. It’s off nlp.seas.harvard.edu.
58.
by nborwankar 3y ago
Yes but to get all the raw data into a form that can be queried by SQL or getting a tool that can query JSON, CSV … via SQL is the non trivial part. Not to mention reconciling date formats etc from multiple sources into one format. 80% of
59.
by nborwankar 3y ago
IMHO it may be more accurate to say “the US has more fluent speakers of American English than any other country in the world.”
60.
by nborwankar 3y ago
Yes

More ›