dimatura

15 ms·

151.
by dimatura 3y ago
I don't have experience with text/nlp problems, but some degree of automation/assistance in labeling is a fairly common practice in computer vision. If you have a certain task where the ML model gets you 90% there, then you c
152.
by dimatura 3y ago
I think that's an insightful point. Personally, I believe an important part of ML architecture engineering is figuring out architectures that make giving "the wrong answer" to a problem impossible, or at least difficult. But
153.
by dimatura 3y ago
The OP-1 is certainly a pleasure to look at. I'd love to have one too, based at least partly on looks, but thinking more pragmatically, I don't think I'd really get the price tag's worth out of it as a music instrument,
154.
by dimatura 3y ago
There's no denying the OP-1 looms large in the world of modern hardware synths. Everyone knows it, a lot of people have one despite the price tag. Its design aesthetic and portability have certainly been influential. How many people ac
155.
by dimatura 3y ago
CoreML has various models already built-in, although as black boxes that accomplish some task like OCR or rectangle detection. There's also a "feature print" model which I believe are intended to be used as hard-coded feature
156.
by dimatura 3y ago
It's true that these functions are in some sense very different to traditional hash functions, because they are designed to create "collisions" (or at least be similar) for elements that are similar. But I still think it is i
157.
by dimatura 3y ago
Yeah, hashing for spatial data is a fairly common technique (and not invented for that paper). It totally makes sense for applications where there is a need to index sparse data in a potentially very large or even unbounded physical space.
158.
by dimatura 3y ago
The main one was https://github.com/JaidedAI/EasyOCR , mostly because, as promised, it was pretty easy to use, and uses pytorch (which I preferred in case I wanted to tweak it). It has been updated since, but at the tim
159.
by dimatura 3y ago
As someone who is better at coding than pen & paper math, I definitely enjoyed seeing a lot more cases where the only practical solution was numerical.
160.
by dimatura 3y ago
When did you do this test? I don't have any numbers handy, but a couple years ago I compared google's OCR vs AWS's on "text in the wild" pictures. AWS' wasn't bad, but it was definitely outperformed by the
161.
by dimatura 3y ago
I've used htmx to write some relatively simple internal tools, and it's been really useful for that. I know a little javascript, but my web skills are really outdated - I acquired them back when you rendered HTML on the server and
162.
by dimatura 3y ago
This is very cool. I remember during my PhD I briefly investigated using these 3D models for synthetic data generation, but there was no straightforward, non-shady way of going about it.
163.
by dimatura 3y ago
I wouldn't be so harsh. The author of the blog post is also the author of the paper. If I had to guess, the author decided (or was advised) that writing a blog post about the paper would increase its visibility (which turned out to be
164.
by dimatura 3y ago
Just wait until they find your wife's phone number. (OK, this wasn't datadog, but some similarly aggressive headhunting agency went from calling me to calling my wife - who is in no way affiliated with my company - to ask her if s
165.
by dimatura 3y ago
This book is a good intro to Lisp itself, which is worth knowing in any case- and symbolic AI is probably one of the best domains for it! As for GOFAI in the age of DL/LLM, yes, you should know it, for a couple of reasons. A lot of the
166.
by dimatura 3y ago
This book is fantastic, it was a great intro to both lisp and (GOF)AI in undergrad. My print copy was pricey - I'm glad to see it available for free!
167.
by dimatura 4y ago
Certainly an improvement on the ol' "cat /dev/urandom > /dev/audio"!
168.
by dimatura 4y ago
As far as I can tell, it can. The code itself has a `LICENSE` file with the Apache license, and the readme says "The model is licensed under the Apache 2.0 license.". Strangely, the FAQ in the blog post doesn't address this q
169.
by dimatura 4y ago
The network architecture and scale don't seem to be a big departure from recent SOTA, but a pretty massive amount of labeled data went into it. And it seems to work pretty well! The browser demo is great. This will probably see a lot o
170.
by dimatura 4y ago
At what point are we allowed to use the "reason" or "learn" for AI then? Can we use these terms for non-human animals? If so, which ones? I think LLMs are at the point where adopting an "intentional stance" (u
171.
by dimatura 4y ago
I eventually ended up switching fields to ML (and that's my current day job), but I started out as an undergrad studying psychology/cognitive science. During those years I started a research project on what is called the "fee
172.
by dimatura 4y ago
There has been work on using AI for this at various levels - at the neural architecture level (finding neural architectures with high throughput/latency for a given hardware), at the algorithm level (finding faster matrix multiplicatio
173.
by dimatura 4y ago
People are definitely building at a high pace, but for what it's worth, this isn't the first work to tackle this problem, as you can see from the references. The results are impressive though!
174.
by dimatura 4y ago
Good question, it's not that easy to find existing words that match that well. If you asked me to do this I'd probably try making up fake words that in my mind, at least would sound closer (and that is also non-trivial given the w
175.
by dimatura 4y ago
Like many current LLM results, they're impressive considering the SOTA of ML until recently, but not very good in comparison to a good human baseline.
176.
by dimatura 4y ago
It would be interesting to see how this compares to other solutions in this space, such as s3fs (the FUSE driver, not the python package), goofys, and the rclone mount feature, among others. This certainly has less features (notably, mounts
177.
by dimatura 4y ago
Yeah, not great, these pronunciation hints kind of sound like what a naive english speaker would guess just from looking at the words (and similarly, like a beginner speaker with a heavy english accent).
178.
by dimatura 4y ago
At least according to this tool, "rock stars" are still fair game. EDIT: never mind, I got it wrong. No rock stars either.
179.
by dimatura 4y ago
Interesting article touching on the meaning of the "digital" in digital computers, and beyond that, how digital computation relates to thought (and even, what is "computation"?). My personal guess is that the Church-Turi
180.
by dimatura 4y ago
To paraphrase Marx (and later, Berman), every concept is pregnant with its opposite.

More ›