facu17y

7 ms·

1.
by facu17y 2y ago
How can they speak of Safety when they are based partly in a colonialist settler entity that is committing a genocide and wanting to exterminate the indigenous population to make room for the Greater Zionist State. I don't do business
2.
by facu17y 3y ago
What's the performance penalty of LoRA?
3.
by facu17y 3y ago
It is legal. Fair use. People have been doing it for ages. Almost every article you've ever read has some fair use of another article, book or news item, etc.
4.
by facu17y 3y ago
Browser vendors are (or should be) managing the abstractions for their own needs, with developer needs expected to be met by framework/library developers. Who says web components are meant for use directly by the developer? Maybe they&
5.
by facu17y 3y ago
Well, if it's so useless why is it on the HN front page? Are there "PR" companies behind promoting items to the HN front page? I'm sure there are because sometimes an article like this comes up at #3 and everyone says it
6.
by facu17y 3y ago
"Four years ago, Altman’s mentor, Y Combinator founder Paul Graham, flew from the United Kingdom to San Francisco to give his protégé the boot, according to three people familiar with the incident, which has not been previously reporte
7.
by facu17y 3y ago
Sam didn't create the breakthroughs behind the current GPT. He did not create the breakthroughs behind the next GPT. None of the people that may follow have the same handle on the tech as Ilya. I mean they built up Ilya's image in
8.
by facu17y 3y ago
You are assuming he wouldn't steal t from OpenAI. He could have a low level employee steal it, and manage to keep it a secret until AGI is born then he takes over the world.
9.
by facu17y 3y ago
What did Sam Altman hide from his board that caused his firing as CEO of OpenAI? 1) That LLMs cannot generalize outside of _patterns_ they pick up during training? (as shown by a recent paper from Google, and as many of us know from our wor
10.
by facu17y 3y ago
Repo has been deleted? I get a 404. I did see it earlier on. I fed the equation image (screenshot at the right frame from their gif then cropped) into ChatGPT (GPT4-V) and it correctly deciphered the equation and gave the correct LaText cod
11.
by facu17y 3y ago
"Despite the increasing complexity and capabilities of machine learning models, they still lack what is commonly understood as "agency." They don't have desires, intentions, or the ability to form goals. They operate und
12.
by facu17y 3y ago
I meant that saying "something is inconsistent sometimes" is weird because inconsistency implies "sometimes"
13.
by facu17y 3y ago
"We do have issues with consistency sometimes" That's a strange statement. Having issues with consistency means that sometimes the output is wrong. What does it mean to have issues with consistency sometimes ? You're
14.
by facu17y 3y ago
Mojo was mature enough for a person I know in the community who ported their Python port of llama2 to it. Also, others pointed out other languages they would rather use. The rationalization in your response obfuscate the real reason you wer
15.
by facu17y 3y ago
This downvoting to -3 is illustrative of how HN down votes are about territorial warfare, ego, etc. Has nothing to do with logic. You don't like to port your python code to nogil python? I'm gonna downvote you. WTF is wrong with y
16.
by facu17y 3y ago
I'd rather port my Python code to Mojo and get, multi-threading, SIMD and other speedups
17.
by facu17y 3y ago
no github?
18.
by facu17y 3y ago
Down-voted because...?
19.
by facu17y 3y ago
What does that say about OpenAI losing money on ChatGPT Plus esp now with multi-modal which I can only assume is very expensive.
20.
by facu17y 3y ago
seems that the course doesn't include Google's highly efficiemt "step by step distillation" method, which was discussed on HN yesterday
21.
by facu17y 3y ago
I had the Arabic translated edition of that book! same fond memories!
22.
by facu17y 3y ago
"to replace GPT-3.5/4" Very inflated statement when it comes to GPT4 since it is a MoE model with 8 separate models each an expert in one area, and you can't replace all 8 models with one model trained for $19. I call BS
23.
by facu17y 3y ago
Every now and then GPT4 outputs a wrong answer. It's impossible to build a reliable product on top of GPT4 that is not a simple chat bot.
24.
by facu17y 3y ago
It might be trained on this question or a variant of it.
25.
by facu17y 3y ago
Except Signal's founder probably has/had a connection with the NSA. All security is for making it hard for the common attacker, and hostile countries. The NSA, most likely, has social engineered its way into every stack and every
26.
by facu17y 3y ago
I think FedNow can serve as the inter-bank settlement rails for CBDC, or the wholesale side of CBDC
27.
by facu17y 3y ago
If we have the budget for pre-training an LLM the architecture itself is a commodity, so what does llama2 add here? It's all the pre-training that we look to bigCo to do which can cost millions of dollars for the biggest models. Llama2
28.
by facu17y 3y ago
the base model may have been but not necessarily the RLHF fine-tuned layers they might have added or the shortcut they're taking during inference due to such fine tuning (or for perf optimization unrelated to fine tuning.)
29.
by facu17y 3y ago
if you have to resort to archaic writing to prove a point about the latest and most advanced piece of technology, ... you're practicing religion, not science.
30.
by facu17y 3y ago
Show me how I can kill you with my LLM, or my GAN or Diffusion Model. I dare you!

More ›