usgroup

20 ms·

301.
Defence Intelligence – Communicating Probability(gov.uk)
17 points by usgroup 3y ago|1 comments
302.
by usgroup 3y ago
Best of luck in what follows! Clojure was the first Lisp I had ever learned. It was really easy to pick up and be productive with. It was data science I tried to use it for where it quickly runs into a myriad headwinds but it is a beautiful
303.
An optimal-stopping quant riddle(emiruz.com)
2 points by usgroup 3y ago|0 comments
304.
by usgroup 3y ago
Bare in mind that this approach is very limited. Any query requiring global awareness of the context will likely be disappointing. For example, "how was the experimental design different compared to how we've previously done this
305.
by usgroup 3y ago
I think any such systems -- at the moment -- will try to avoid having to fine-tune the completion model, which leads to this kind of solutions: 1. Semantically index your documents. 2. Given a prompt, extract relevant paragraphs from your o
306.
by usgroup 3y ago
Yeah I thought the same -- it struck me at first blush as if it was some kind of super simple architecture that didn't use transformers, and then in the diagram i saw they used BERT to produce the embeddings!
307.
by usgroup 3y ago
That "correlation is not causation" is obvious to researchers but they cannot pick the treatments in their study, so instead what they have to do is try to remove confounding factors instead. E.g. "people that play chess may
308.
by usgroup 3y ago
It wasn't obvious to me why the authors chose kNN for the classifier. Since they produce a distance matrix, they could have used multi-dimensional scaling to convert the matrix to factors, and then used a tree algorithm such as xgboost
309.
by usgroup 3y ago
Good old PHP —- still making the front page in 2023.
310.
LLZ – LLM via Lempel-Ziv Compression(bactra.org)
3 points by usgroup 3y ago|0 comments
311.
by usgroup 3y ago
its not a dumb question -- optimal coding in theory assigns codes of shorter length to more frequently occurring tokens hence by design longer codes imply lower probabilities: that's how compression is achieved.
312.
by usgroup 3y ago
For anyone else wondering how this works out: 1. You want: p(x) ~ K^(-|x|), where K=256. 2. log p(x) ~ log K^(-|x|) = -|x| log K 3. he is using log(softmax) ~ log(e^x) 4. and log(e^(-|x| log(K))) = -|x|*log K as required.
313.
by usgroup 3y ago
i suggest you get all the unique characters from the moby dick text and use it as the alphabet if you're generating from it, or truncate the text to just your hand selected characters else gzip cannot even approximate an optimal code i
314.
by usgroup 3y ago
Hey author, would you mind explaining the conversion from compression length to probability please? Namely this line: scipy.special.log_softmax(-code_lengths self.conversion (1/temperature)) Your codes are K-ary but this doesn't l
315.
by usgroup 3y ago
i am curious to know the results when using a text-specific compression method. Text specific compression would more likely approximate Kolmogorov complexity for their use case. I'm also curious whether a lossy text compression scheme
316.
by usgroup 3y ago
A related question -- when fine tuning a model like this to a specific corpus, how does the fine tuning effect the actual chat capability, since the chat model weights seem to come as a separate model? Does one fine tune the LLM+Chat model
317.
Estimating gym goers using mark-recapture experiments(emiruz.com)
1 points by usgroup 3y ago|0 comments
318.
Estimating gym goers using mark-recapture experiments(emiruz.com)
1 points by usgroup 3y ago|0 comments
319.
by usgroup 3y ago
I guess you know how bad your concussion was and its prognosis, but having post-concussive symptoms that last <= 18 months is not uncommon. I'm not even sure that the post-concussive symptoms are directly related to prognosis. I und
320.
by usgroup 3y ago
Yeah i built a house pricing web service in Rust under the Axum framework. It is seriously fast, but in my opinion it a very difficult choice for web if speed is not the essential requirement. I feel -- at least at my level of Rust competen
321.
Blocking, covariate adjustment and optimal experiment design(emiruz.com)
1 points by usgroup 3y ago|0 comments
322.
by usgroup 3y ago
> Warren Buffett argues that active investing doesn't make sense because it loses out to passive investing in the long run. I agree with this. He is an active investor… He may argue for buy and hold the S&P for ordinary folks, b
323.
by usgroup 3y ago
I wasn’t commenting on how difficult it is to be hired —- that’s relative to many things. But in terms of the least you need, I think I’m unfortunately right because I have seen it many times over the least few years. We will see over the n
324.
by usgroup 3y ago
I think studying what happened to the “data engineer” role is a good indicator. There was a brief moment in time wherein data engineers were computer scientists specialised in distributed systems and data processing algorithms on commodity
325.
by usgroup 3y ago
If you think it’s due to remote, apply for some non remote stuff and test your hypothesis. If you think it might be salary expectations, test your hypothesis by asking for a lower salary. And so on. If you’re gonna spend 6 months getting a
326.
by usgroup 3y ago
“ The fixed sort solutions for sort 3, sort 4 and sort 5 discovered by AlphaDev have been integrated into the standard sort function in the LLVM standard C++ library3.” Not exactly what the title would lead you to believe.
327.
by usgroup 3y ago
I think user-in-the-loop use cases are not uncommon for semi-supervised learning broadly speaking. In your case I'm not sure you need to conceive anything as 3-valued logic. For example: 1. There is either a tag or there is not. 2. If
328.
Automatic Differentiation in Prolog(arxiv.org)
2 points by usgroup 3y ago|0 comments
329.
Semi-supervised clustering with logic programming (Prolog)(emiruz.com)
5 points by usgroup 3y ago|2 comments
330.
by usgroup 3y ago
Its true for logic programming languages too (e.g. Prolog, Picat, Mercury, etc), so I do not think its to do with line noise languages per say nor a lack of examples (in the case of Prolog). It'll write it but it treats it like Python

More ›