LarsDu88

37 ms·

691.
by LarsDu88 2y ago
This is simply not true. They are building giant fabs in Arizona and Ohio from those subsidies. It takes YEARS to realize the benefits of these changes. The problem is, to stopgap TSMCs growing lead, Intel resorted to cranking up voltage on
692.
by LarsDu88 2y ago
I mean, for all of Amazon's business intelligence, running on servers operated by Amazon... tis a mere pittance.
693.
by LarsDu88 2y ago
As someone who used Ray for ML, this is impressive as hell. I did not even realize it was that viable a substitute for straight data processing tasks.
694.
by LarsDu88 2y ago
Wow that is not NEARLY as expensive as I would have imagined considering the scale of the data involved.
695.
by LarsDu88 2y ago
This is an insane statment. There's a reason Nvidia has been able to coast from one boom (crypto) to another (ai), and has bullets in the magazine for several other moonshots that are coming to fruition (autonomous vehicles; finally ro
696.
by LarsDu88 2y ago
Does AMD have a library like TensorRT that optimizes inference with operator fusions and kernel optimizations though? TensorRT can halve the inference compute for certain Nvidia inference workloads
697.
by LarsDu88 2y ago
Rus is what these people called themselves. They were folks lured to that part of the Mediterranean by promise of work in the Byzantine military as Varangian Guards. A subset of these Norsemen sailed up the Dniper to what is today Kiev and
698.
by LarsDu88 2y ago
This does seem like something ai could automate
699.
by LarsDu88 2y ago
Well the point is, both LoRA and Dreambooth require fine tuning the model (i.e. training)
700.
by LarsDu88 2y ago
OP here: Thanks for changing the title as well!
701.
by LarsDu88 2y ago
Hmmm, that'd be an interesting startup idea! How do ad blockers work exactly?
702.
by LarsDu88 2y ago
This is a much bigger thing than the llama3.1 release. Llama 3.1 doesn't really help Meta's bottom line. But content creation and ads are Meta's killer app. By having a model that doesn't require finetuning, they just ch
703.
by LarsDu88 2y ago
Up until recently, to insert yourself into an image generation algorithm, you had to use a technique like Dreambooth, which involves finetuning the model itself with a new mapping of the subject to a rare token. Meta just released and produ
704.
Tuning-Free Personalized Image Generation(ai.meta.com)
82 points by LarsDu88 2y ago|45 comments
705.
by LarsDu88 2y ago
Meta is also trying to commoditize AR hardware interestingly enough by releasing their Quest operating system to third parties like samsung. I could write a whole nother article on that phenomenon as I have developed VR apps myself.
706.
by LarsDu88 2y ago
Author here: For cloud hosts like Amazon, Google, and Microsoft, the product is their inference hardware (mostly GPUs). The bigger the open model, the better! For Meta however, its a bit more mysterious. From what I can glean, the stated ob
707.
by LarsDu88 2y ago
Damn this autocomploot
708.
by LarsDu88 2y ago
Author here: The title "Why big companies like Meta want to commoditize open source and open weight models to increase demand for their complimentary services" did not quite have the same ring to it to be quite honest
709.
by LarsDu88 2y ago
Author here. I had a brain farther when posting this to Hackernews. It should have Why in the posting title
710.
by LarsDu88 2y ago
My musings on the recent commoditization of large scale LLMs
711.
Big tech wants to make AI cost nothing(dublog.net)
88 points by LarsDu88 2y ago|80 comments
712.
by LarsDu88 2y ago
Obligatory reminder of why tech companies subsidize open source projects: https://www.joelonsoftware.com/2002/06/12/strategy-letter-v/
713.
by LarsDu88 2y ago
Source?
714.
by LarsDu88 2y ago
SpaceX doesn't really do any sort of frontier research. They do technology. They are taking concepts that have been done in the past and doing them more economically. They have pushed the state-of-the art with projects like Raptor-2, b
715.
by LarsDu88 2y ago
That one's on Intel. They marketed Optane poorly and created a subpar product for the market they were tackling. The whole point of memristors is to make something better than DRAM and SRAM. Perhaps even blow up the dichotomy between s
716.
by LarsDu88 2y ago
Just like cough Bell Labs maybe?
717.
by LarsDu88 2y ago
Google already did this. Google X, Google Brain, Deepmind... The invention of the transformer architecture that underpins modern AI, dozens of core papers in the field. What they haven't revolutionized is materials science. Memristor t
718.
by LarsDu88 2y ago
I'm on my 5th dose, and I now have infrared vision and can smell colors. Feelz great man!
719.
by LarsDu88 2y ago
Lets see how the market caps of these two companies compare to the GDP of those countries... Oh, greater than all of them combined you say? Several times more if you exclude Nigeria and Ireland And a big chunk of Ireland's GDP is reall
720.
by LarsDu88 2y ago
Understanding the awesomeness of Blender is when I realized that the fate of all software is to be commoditized. It is almost an anomaly that the highest market cap companies are software companies until you realize that most of their value

More ›