GaggiX

25 ms·

511.
by GaggiX 2y ago
if the comment is flagged then it should be visible only if you have "showdead" on on your account, so I don't see how it can be crawlable, same for the comment thread under the flagged comment.
512.
by GaggiX 2y ago
Many of the actual physicists out there are ML researchers.
513.
by GaggiX 2y ago
Most of my entertainment is watching dudes sitting in their chairs talking into a microphone. I find it more entertaining than the billion dollar entertainment industry.
514.
by GaggiX 2y ago
I wonder how much compute was required to remove the distillation.
515.
by GaggiX 2y ago
Flux Pro (v1 and v1.1) is a close source model.
516.
by GaggiX 2y ago
Have you run Flux Pro offline?
517.
by GaggiX 2y ago
Have you tried Ideogram v2?
518.
by GaggiX 2y ago
Now we know what blueberry was, I thought it was some new model from OpenAI. There is still a mysterious blueberry on a image to 3D model arena: https://huggingface.co/spaces/dylanebert/3d-arena Hopefully this mod
519.
by GaggiX 2y ago
I use NovelAI and that's also profitable. I would be surprised if Elevenlabs wasn't profitable right now.
520.
by GaggiX 2y ago
That's not true, there are plenty of companies that make a profit, Midjourney, for example, an obvious one.
521.
by GaggiX 2y ago
Well but it actually doesn't, unless you're looking only at MMMU.
522.
by GaggiX 2y ago
Details will be shared tomorrow, but from what I have read they have distilled the large model decoder into this turbo that only has 4 layers instead of 32, the encoder should remain the same size. Similar to https://github.com&#
523.
New OpenAI Whisper model: "turbo"(github.com)
13 points by GaggiX 2y ago|1 comments
524.
by GaggiX 2y ago
They are way bigger than 1B or 3B.
525.
by GaggiX 2y ago
The fact that LLMs are usually trained on tokens and not on characters, doesn't really speak about what generative AI is going to reach or not. >most people selling LLM today pretend it is AGI Who are these "most people"?
526.
by GaggiX 2y ago
Yes, I agree, for these small models it's wasted potential to be closed source, they can only be used effectively if they are open. EDIT: HN is rate-limiting me so I will reply here: In my opinion 1B and 3B truly shine on edge devices,
527.
by GaggiX 2y ago
People use these models for different reasons.
528.
by GaggiX 2y ago
That is indeed their claim, but if you read the linked page you will see that it is not their effective token window.
529.
by GaggiX 2y ago
No one listens to an hour of actual vomit just because a friend sent it to them, you should value your time more if you do at minimum.
530.
by GaggiX 2y ago
Even if you think your friend would only send good things, you would realize that something is vomit in less than an hour. I cannot understand someone listening to something for an entire hour and then whining that they waste their entire h
531.
by GaggiX 2y ago
If it was vomit, it will be recognized quickly, AI or not, not an hour of listening for sure; yes, even if it was sent by a friend.
532.
by GaggiX 2y ago
3-4 hours a day for a year is not even realistic, unless it's a language that already has a lot in common with yours, like Italian and Spanish.
533.
by GaggiX 2y ago
If it was vomit, why did you spend an hour on it? People complain about 2 minutes of audio sometimes, I cannot imagine a full hour of an unknown podcast, it must have been quite interesting.
534.
by GaggiX 2y ago
It already feels more nuanced than the usual podcast.
535.
by GaggiX 2y ago
I hope they evacuate the crew and bomb it in the middle of the ocean so we can film the explosion in 4K. Unfortunately, I don't think that's going to happen.
536.
by GaggiX 2y ago
Let's not ignore better architectures, training techniques and computing power.
537.
by GaggiX 2y ago
You can generate AI images that do not have the "AI look": https://ideogram.ai/assets/image/lossless/response/icQM0yZQQ... And it's been two years since SD v1, a model that was not able to
538.
by GaggiX 2y ago
They are not even that bad anymore to be honest.
539.
by GaggiX 2y ago
Machines were not able to deal with non-formal problems.
540.
by GaggiX 2y ago
Ideogram v2 first take on the prompt: https://ideogram.ai/assets/image/lossless/response/v_LgyXI1Q... https://ideogram.ai/assets/image/lossless/response/V4RRDJZJS...

More ›