3abiton

40 ms·

661.
by 3abiton 3y ago
Proxy kpis, like how many tech YouTubers are pivoting to the "check my home server you guys", and the popularity of home server desktop hardware (on the shopping sites I follow)
662.
by 3abiton 3y ago
Great follow-up. It is odd to have this inconsistency in behaviours, and a bit scammy from openai, as these are paid features.
663.
by 3abiton 3y ago
Same, CodeLlama 70B is known to suck. Deepseek is the best for coding so far in my experience, Mixtral 8x7B is another great contender (to be frank, for most tasks). Miqu is making a buzz, but so far I haven't tested it personally yet.
664.
by 3abiton 3y ago
But there are decent OSS alternatives that can rival it's functionality, so why go with Tableau and it's baggage?
665.
by 3abiton 3y ago
If you look right now at the homeserver market, it's getting larger and larger. With APUs, it could be a great bargain for a self-hosted PC console with decent steam deck like performance. Linux is getting spoiled with steam compatibil
666.
by 3abiton 3y ago
This is a deal breaker for many HN users, including me. Especially a piece of software with such privileges.
667.
by 3abiton 3y ago
Thanks for sharing, it's so exciting to see so many OSS BI frameworks
668.
by 3abiton 3y ago
GPT-blinkist
669.
by 3abiton 3y ago
ZLUDA has its own baggage for now. Still lacking many support, but that changing fast, maybe in a month or so.
670.
by 3abiton 3y ago
So odd they call GPT4+Bing a "RAG' system
671.
by 3abiton 3y ago
What you describe as annoying about your experience sounds like a case of stakeholder management rather than frustration with programming. But I totally agree with you too.
672.
by 3abiton 3y ago
It would be interesting to see the comparison between Prolog and a similar purposes LLM
673.
by 3abiton 3y ago
I have wasted so many nights debugging and trying to figure out how to get LLMs to run on my AMD iGPU. Despite pytorch and xformers (unofficially) being supported, I was still unsuccessful. Cuda on the other hand, it just works. I don&#x27
674.
by 3abiton 3y ago
But this is not new, there have been decades of effort in this field (countering the dark energy theory), and in my Uni's physics department we always had the QT guy vs the DE guy trying to convert students to their legions.
675.
by 3abiton 3y ago
This will be worse will llms
676.
by 3abiton 3y ago
It only takes one successful attempt, and it's worth it
677.
by 3abiton 3y ago
That's one way to go. Yolo on a prank.
678.
by 3abiton 3y ago
Nonetheless, this is extremely exciting, unlike RWKV and Retention Network
679.
by 3abiton 3y ago
I think is the key, most likely it was a big part of their training dataset.
680.
by 3abiton 3y ago
To piggyback on your point, some people like to be exploited, and I met so many hard working people who never realized the value of their work and how important they were to their company, maybe just blinded by the culture, but were severel
681.
by 3abiton 3y ago
It's all about the long tail cases.
682.
by 3abiton 3y ago
Tesla P40
683.
by 3abiton 3y ago
Are there any MoE multimodal approaches?
684.
by 3abiton 3y ago
Agents and transformers are kinda different things, built on different layers.
685.
by 3abiton 3y ago
Worth reading quality land. It's about that last point you made.
686.
by 3abiton 3y ago
His previous video onLLM tramsformer foundation is extremely useful.
687.
by 3abiton 3y ago
How does it compare to selenium or puppeteer?
688.
by 3abiton 3y ago
Never heard of it, what an interesting read
689.
by 3abiton 3y ago
It read like a teenager wishlist essay, less technical more whimsical.
690.
by 3abiton 3y ago
Also reddit exodus, aftee their api changes. Don't understimate people's imcline for good open source (or 3rd party clients)

More ›