resters

27 ms·

331.
by resters 2y ago
any benchmarks vs phi-3?
332.
by resters 2y ago
It’s true that current models do not do formal reasoning, my point is that it is possible to use tokenization to do it. See my comment in the other thread.
333.
by resters 2y ago
I'm working on this: Abstract: This paper presents a novel framework for multi-stream tokenization , which extends traditional NLP tokenization by generating simultaneous, multi-layered token representations that integrate subword em
334.
by resters 2y ago
I realize there is subtlety to the question of which is first. An infant, crying when it is hungry and pre-linguistic, is applying modus ponens. C -> F crying implies food, so I cry and then I get fed. Language grows in humans just like
335.
by resters 2y ago
I think models could be designed that in separate layers created "logical system" representations which could feed back into the output, much like how attention works. Attention is about relevance, the logical layers could be base
336.
by resters 2y ago
I don't disagree, however I'm optimistic because most of the current reasoning "ability" of LLMs comes from the accidental reasoning embedded in language patterns. For example, the prompt completion: "The mouse has
337.
by resters 2y ago
I think it's obvious that LLMs will be able to do "reasoning" far better than humans. We must separate our notion of what is remarkably human. Rarely is it the reasoning, it's the intuition that a logical path exists --
338.
by resters 2y ago
I'm not making excuses for Russia's treatment of civilians at all. However the US killed a lot of Japanese civilians during WW2 and is considered "civilized" by some today. I think the geopolitical view offers a lens thr
339.
by resters 2y ago
Chechnya was hugely important to Russia geopolitically and so it should not have been a surprise that it would not want to let Chechnya become independent. Russia already had a mix of rivals both for political Islamist reasons and oil/
340.
by resters 2y ago
I read the letter and it sounded like the "Bush Doctrine" of preemptive aggression. The claim is that Russia is by nature aggressive (based on a very long-term historical trend that does not include its current regime) and so agg
341.
by resters 2y ago
I tried to download my raw data recently and it took days. Seems like a lot of customers are trying to download it and cancel after the turmoil. I think 23andme has always been held hostage by its scientists who have stopped it from offeri
342.
by resters 2y ago
I think if there is one "value" that stands out from Google's culture (as it is reflected onto its customers) it is tremendous lack of empathy. - Google maps appears to be designed by non-drivers. Much of hte time it is impos
343.
by resters 2y ago
Not in major US cities such as where I live.
344.
by resters 2y ago
I had done this a few years earlier bc USPS kept damaging heavy items, but it is apparently no longer possible. For a while things got delivered via Amazon's subcontracted trucks but that stopped and now it's just whatever level o
345.
by resters 2y ago
I care less who delivers it than whether it actually arrives in <= 2 days. In my zip code there are various blocks (including my own) where virtually nothing gets delivered in <5 days. This is in one of the biggest cities in the US.
346.
by resters 2y ago
Amazon currently does not honor its prime shipping agreement with all domestic customers and has no customer service path to get the issue resolved. In some places USPS is unreliable and Amazon insists on using it repeatedly, violating its
347.
by resters 2y ago
YC gave Palantir a massive platform at startup school to present the company and vision and recruit talent. Far more than the seed investment would have been worth. With respect to the ares industries, YC buys into the neocon view strongly
348.
by resters 2y ago
Some really great ideas came out of Rails 1 and 2. After that Rails became more of a business, but good ideas still came out of the ecosystem and germinated (in my view more successfully) in the fertile soil of node / js, as well as s
349.
by resters 2y ago
If anyone needs to see what happen, read Marc Andreesen's cringy posts on X. The guy is a MAGA promoter and sounds more and more like an elderly "the country is going to hell in a handbasket" geriatric conservative every sec
350.
by resters 2y ago
True, but why pay for the obsolete tech in the first place? A simple amplified speaker system and slot for a tablet would cost a lot less and avoid much of the obsolescence one faces with cars. Oops, the obsolescence is a key part of the p
351.
by resters 2y ago
I’ve wondered for a while why cars don’t simply ship with a tablet sized indentation, and adapters for popular tablets to connect an amplified speaker system. It appears the reason is because Congress thinks it knows best what type of audi
352.
by resters 2y ago
sometimes one must give it current docs to avoid that. Ironically gpt-4 always tries to use the old chat completions api when writing code to consume the open-ai api.
353.
by resters 2y ago
give me an example of the kind of thing you are curious about how to build in this way, and I will give you a step by step.
354.
by resters 2y ago
For types of code that LLMs are "good at", the results are quite good if the LLM is told pretty much how to solve the problem. Generally they can map English language text into working code modules, can write tests for the modules
355.
by resters 2y ago
I think it's already surpassed 10x, and is closer to 25x or 30x. You just have to know what to ask it to do.
356.
by resters 2y ago
AI (LLMs in this case) reduce the value of human conscientiousness, memory, and verbal and quantitative fluency dramatically. So what's left for humans? We very likely won't have as many human software testers or software engineer
357.
by resters 2y ago
This is great! Does anyone know if the llama models are trained to do function calling like openAI models are? And/or are there any function calling training datasets?
358.
by resters 2y ago
Maybe Google is holding back it's far superior, truly sentient AI until other companies have broken the ice. Not long ago there was a Google AI engineer who rage quit over Google's treatment of sentienet AI.
359.
by resters 2y ago
> finding actual citizens that could fill these jobs and build their careers. So then let's let them immigrate and gain citizenship!
360.
by resters 2y ago
China probably measures it. In the US it's probably driven by a Fox News opinion poll of rural PA voters.

More ›