lossolo

64 ms·

931.
by lossolo 4y ago
> platforms will follow and improve their technology for recognizing spam Written language has patterns, and your response and mine also follow these patterns. Language models like GPT-3 and ChatGPT can learn these patterns because there
932.
by lossolo 4y ago
TSLA -73.82 (-40.36%) past month only
933.
by lossolo 4y ago
And Google developed LaMDA model (superior to chatGPT) a lot of time before OpenAI, additionally without their research into transformers and publishing results, OpenAI would probably not even have ChatGPT now/yet.
934.
by lossolo 4y ago
From first google result: > TSLAQ (pronounced "Tesla Q") is a loose, international collective of largely anonymous short-sellers, skeptics, and researchers who openly criticize Tesla
935.
by lossolo 4y ago
It's 1 billion model with Fp16 precision so 4-6 GB max.
936.
by lossolo 4y ago
I've been using it more and more at work, and it's already saved me hours by generating bash commands and simple scripts/servers that I would otherwise have to search for on Google and adjust to my specific use case from mult
937.
by lossolo 4y ago
For all models smaller than 62B, direct prompting outperforms CoT. The first model where CoT outperforms direct prompting is Flan-cont-PaLM 62B on BBH. For 540B models, there are more settings where CoT outperforms direct prompting, but not
938.
by lossolo 4y ago
> The European Central Bank has arguably failed its mandate of keeping price stability within the eurozone when some countries have more than 20% inflation. It's not that simple. They could rise interest rates more but Italy and Gre
939.
by lossolo 4y ago
> scientific first principles type of thinking Twitter CEO Elon Musk blasts Rep. Adam Schiff in deleted tweet: 'Your brain is too small'[1] 1. https://www.foxbusiness.com/politics/twitter-ceo-elon-musk-b...
940.
by lossolo 4y ago
Seems like very critical article about Amazon in Post: https://www.washingtonpost.com/technology/interactive/2022/a...
941.
by lossolo 4y ago
> I think her value add is very clearly "she is famous, people will buy shit that she puts her name on". Musk example[1] of this, he sold 1 million USD worth of perfume with smell of burnt hair in a few hours. https:/&#x2
942.
by lossolo 4y ago
> Most likely it's going to be very rough. I don't know why are you drawing such conclusion, in all democracies still people decide by voting, if most people will lose their jobs then what do you think they will do? Vote to be
943.
by lossolo 4y ago
Sure but you can only verify facts like "when was <someone> born?", you can verify this today easily with knowledge database but that's not what is interesting in ChatGPT, what's interesting is what it can generate
944.
by lossolo 4y ago
For anyone interested it's called retrieval transformer. Here is example of one from Google: RETRO[1][2] 1. https://arxiv.org/abs/2112.04426 2. https://jalammar.github.io/illustrated-retrieval-tran
945.
by lossolo 4y ago
> Sure there will always be something we'd want to control but we can go in more calm, clearheaded way towards it rather than just use force because we have the biggest stick. Unfortunately no, you can't go more calm because pr
946.
by lossolo 4y ago
> We wouldn't have to fight those wars for resources. You mean wars for oil? Fusion would not solve that, that wars were not about energy per se but about control and domination, keeping USD as world reserve currency and US as world
947.
by lossolo 4y ago
Seems like majority of problems in this log are devops problems, which seems to be combination of ML people doing devops work while not having experience with devops work and really bad cloud vendor. I've been running multiple bare met
948.
by lossolo 4y ago
The whole input of your current session is fed into the model, that's why it tricks you that it "KNOWS" like human would know, in reality this is a lot of data, computation and statistics without any reasoning, that's wh
949.
by lossolo 4y ago
ChatGPT: > The upper part of your digestive system has lots of special cells that can taste things and sense fat. These cells have tiny helpers inside them called receptors, and when they taste something sweet, they send a message to you
950.
by lossolo 4y ago
Have you seen RealFlow? I was having fun with their fluid physics simulator a few years ago, it's really good. https://realflow.com/
951.
by lossolo 4y ago
> Google releases research papers but can’t productize anything If you mean actual products and not just open sourcing models and code then: https://cloud.google.com/products/ai They also implement a lot (the most i
952.
by lossolo 4y ago
> if Google offered Search "Premium" using the latest LLMs, how much would people pay for it? I think they need to solve hallucination problem first, they are already working on optimization eg FLAN-T5 (smaller LLM models same
953.
by lossolo 4y ago
> Has Google/Alphabet publicly released any of their AI models yet? You mean NLP field changing models from Google like BERT [1]? or Transformers paper [2]? or T5 model [3] (used by company doing ChatGPT like search currently on the
954.
by lossolo 4y ago
What's funny is that most of this ground breaking LLMs you see now are based on Google published research about transformers, and they have better performing models in house than anything publicly available on the market.
955.
by lossolo 4y ago
11B parameters, so I guess it's FLAN-T5[1]? combined with retrieval transformer (something like RETRO[2])? 1. https://arxiv.org/abs/2210.11416 2. https://arxiv.org/abs/2112.04426
956.
by lossolo 4y ago
Who knows? Maybe we will find a bug in current universe implementation and we will be able to breakout from the simulation in different form.
957.
by lossolo 4y ago
> So how 'big' is GPT-3? For inference so basically running it you need multiple GPUs and hundreds of GBs of GPU memory. As to model size it's around 100x bigger than SD. You can forget about running it locally unless you
958.
by lossolo 4y ago
SD is only 4 GB in size, you can run it on one 8-10 GB GPU, GPT-3 is around 350 GB so around 100x more, 99% of humanity can't run it locally currently even if they wanted.
959.
by lossolo 4y ago
It doesn't have any logic, it's just prediction based on statistics. There is so many examples already floating around that it has no logic but I will give you really simple one from my experiments: I told it to: > curl somedom
960.
by lossolo 4y ago
You are mistaking again model with dataset. To be on the same knowledge depth they both have to use the same dataset, the difference is that in search engine you have points of references, rankings and reputation of sites, human discussion

More ›