pennomi

12 ms·

241.
by pennomi 3y ago
Base model choice is a big one. Probably didn’t pay close attention to their negative keywords either. You can specifically tell models to strongly avoid making nsfw images. Plus most image generation pipelines have an optional secondary st
242.
by pennomi 3y ago
This is like measuring your active calories vs your total calories in a workout. Turns out that a meaningful percentage of calories are burned just by being alive, not from the work that was done.
243.
by pennomi 3y ago
Just playing Devil’s Advocate here, but we assume that AI consuming AI-generated content is going to inherently hurt the next generation of AI training. But I’d argue that good AI content is more valuable training data than bad human conten
244.
by pennomi 3y ago
Is there significant risk in slowly clawing away user experience? If it happens incrementally enough, users are unlikely to leave.
245.
by pennomi 3y ago
It doesn’t affect Nintendo, yet. I suspect that if this legislation is successful, it will expand in scope.
246.
by pennomi 3y ago
Plus SVG can’t properly wrap text, which matters for some use cases.
247.
by pennomi 3y ago
Hell, I’d regularly donate to the OpenAI Crowdsource Fund if it guaranteed their research would be open sourced.
248.
by pennomi 3y ago
Happened already with the SDXL 0.9 weights leak. People started training off of that and it quickly became wasted effort.
249.
by pennomi 3y ago
I have a suspicion that organoids will be to artificial intelligence what ornithopters are to flight. Technically possible, but reinventing it from scratch instead of following biology strictly will give us dramatic performance improvements
250.
by pennomi 3y ago
No, I’m saying that their policies are more inclined to be interpreted as nice/charitable than conservative policies. This covers wide areas such as immigration (allow migrants into the country), welfare (increase social programs, such
251.
by pennomi 3y ago
Surely emphasizing concepts like diversity, equity, and being fair to all people regardless of situation or origin in the training data simply makes the results lean slightly left. Essentially trying to make a “nice” AI will skew left due t
252.
by pennomi 3y ago
I feel like to be truly considered OSI-compatible open source, you’d need three things: 1. Model code license (MIT, GPL, etc.) 2. Model weights license (CC?) 3. Model training code license (MIT, GPL, etc.) I don’t think anything new needs
253.
by pennomi 3y ago
To be fair, a friend I know ran into a string of similar issues when attempting to get her gender changed on her license and passport. The entire system was rife with incorrect advice from workers and broken documentation, which caused seve
254.
by pennomi 3y ago
Why? Because it sells better, or is cheaper to produce, or it lasts longer on shelves. Virtually all food decisions are made with economics, not health, driving them.
255.
by pennomi 3y ago
It’s probably considered enough of a transformation that it no longer falls under the license, but that’s the main question on copyright that’s getting solved in courts right now anyway. I wonder if the Share Alike license could be updated
256.
by pennomi 3y ago
You can pack two trits into three bits, however. So one byte could hold 5 values instead of 4.
257.
by pennomi 3y ago
500 t/s is uncomfortably fast to me. Generating high quality answers at speeds faster than I can read is the point at which I feel like LLMs are magic. I’m glad people are doing it though, and I’ll happily adapt to accessing inference
258.
by pennomi 3y ago
As a concrete example, when Blender finally made left click the default selection button (in harmony with virtually every other 3D software out there) it made the learning burden much, much easier for new users to shoulder. This, combined w
259.
by pennomi 3y ago
Maybe. Until we find there’s a better way to encode the information and need the unfiltered, original context so it can be used with that new method.
260.
by pennomi 3y ago
That being said, Steam Deck probably sells more games than, for example, the Steam Controller did. Probably sells more games than their VR headset did too.
261.
by pennomi 3y ago
A decentralized solution seems ideal for dating and other social apps, since then the incentives would perfectly align with the user’s needs rather than corporate profits.
262.
by pennomi 3y ago
> Granted, those under 40 are still the poorest of the generations. Yep, this title seems to be cherry-picked to sound positive despite not actually changing much.
263.
by pennomi 3y ago
Blink isn’t dangerous, Chrome is. You can have a privacy focused Blink browser.
264.
by pennomi 3y ago
My grandchildren will be begging me to join the hivemind, but my crotchety old brain will complain back at them about how a keyboard and mouse are just fine.
265.
by pennomi 3y ago
Meta basically got a ton of free R&D that directly applies to their model architecture. Their next generation AIs will always benefit from the techniques/processes developed by the clever researchers and hobbyists out there.
266.
by pennomi 3y ago
Allowing users the ability to choose their preferred app store is dumb?
267.
by pennomi 3y ago
I assume the human maintains some of the necessary context in their meat memory.
268.
by pennomi 3y ago
Optimizing AIs to be addictive to humans is always how humanity will end. It was the natural end to social media, and market forces will force the same to happen in this industry. People worry about the robot uprising killing all humans but
269.
by pennomi 3y ago
Absolutely my experience as well. Scientists write code that works, but is a pain to reproduce in any sort of scalable way. However it’s been getting better over time as programming is becoming a less niche skill.
270.
by pennomi 3y ago
I think if you were to comply with all the branding guidelines, you would need to say something along the lines of“Blocks are like LEGO™ bricks …”. Since there’s little chance of the product getting confused with official LEGO products, yo

More ›