make3

26 ms·

721.
by make3 4y ago
By very far most of the work in nlp now uses pretrained models. So people use HuggingFace Transformers now. https://huggingface.co/docs/transformers/main/en/index HuggingFace Transformers is a huge high
722.
by make3 4y ago
are these bad? asking as someone who knows nothing of any of them
723.
by make3 4y ago
the goal is ease of use though, not serious analysis
724.
by make3 4y ago
can't find the model size or a description of the dataset on the text-generator page
725.
by make3 4y ago
can't you just move your eyes
726.
by make3 4y ago
couldn't see those on my phone. Wish I could delete my comment.
727.
by make3 4y ago
I guess the question would be if this is more harmful than a plane of similar pulling power
728.
by make3 4y ago
the video is really not convincing lol, it doesn't look like it's doing much more than gliding from the initial launch force, which is due to an external (moving) force. EDIT: ok this one is more convincing https://m.yo
729.
by make3 4y ago
research show that people are apparently terrible at evaluating how many calories they take, even dieticians. by cutting more addictive food, it's possible that you're actually eating less than your otherwise would, even if the nu
730.
by make3 4y ago
let's be real, this has no chance of seeing any real adoption, except as a novelty or for technical demos, because the runtime takes multiple seconds to download, which is insane. reading that document without anything addressing this
731.
by make3 4y ago
It's widely accepted that OpenAi doesn't release its models to make money from them, not because they really think they would be harmful
732.
by make3 4y ago
sound doesn't hit the spot
733.
by make3 4y ago
With the large performance increases brought by the more recent versions of Python, it's not clear to me that installing this is faster than just upgrading Python. It is more simple though I'll give you that. If it also faster wit
734.
by make3 4y ago
<deleted>
735.
by make3 4y ago
requires creating a separate process? requires passing the memory from one process to the other.. requires a filesystem & requires the security analysis for having production software interact with the file-system?
736.
by make3 4y ago
that's so minor
737.
by make3 4y ago
I mean I don't know how that makes it any better from a reproducibility stand point lol
738.
by make3 4y ago
this feels really unpythonic, would be very surprised if I saw this in code
739.
by make3 4y ago
I'm not sure I see how being asked for sex would be a crime... I'm all for decriminalization though
740.
by make3 4y ago
wtf does this have to do with extortion being illegal. it could be a shakedown for money or whatever too
741.
by make3 4y ago
you seriously thought it was about reinstating all of Instagram?
742.
by make3 4y ago
that's the thing, i wish yaml was just a config format, that it does other weird shit like calling functions is the problem
743.
by make3 4y ago
there are lots of articles on the subject, your can use Google
744.
by make3 4y ago
it says, in a week or so
745.
by make3 4y ago
he's also known for single handedly starting the field of adversarial examples, which is a huge field. he basically started two fields. he's huge
746.
by make3 4y ago
why do you think that this would help performance? a quick read says cosmo is slower than regular libc. maybe it would be more portable, but not faster
747.
by make3 4y ago
I'm not sure it's very relevant to say in a discussion of the answer of "how do we improve Python" is "don't use Python". People have all kinds of valid reasons to use Python. Let's keep this on topic
748.
by make3 4y ago
"too much", if you take small amounts not too often, it's fine
749.
by make3 4y ago
does it not work on WSL ?
750.
by make3 4y ago
this doesn't really sound like the type of workload a laptop should be evaluated on... although it is an extremely strong laptop so without gaming I wonder what the intended workload really is

More ›