trash_cat

16 ms·

91.
by trash_cat 2y ago
Granted I only travel with hand luggage, I never had any issues with cancelled or delayed flighs when booking separate tickets. Even so they are relatvely cheap. It is even cheaper if you book your return flight ticket separately.
92.
by trash_cat 2y ago
They changed the meaning of open from open source to open to use.
93.
by trash_cat 2y ago
Author says that we need to figure out the human in the loop problem as we having a less need for a human here. One could argue this is a good thing about it.
94.
by trash_cat 2y ago
It's not that bad...It's quite easy to follow and understand.
95.
by trash_cat 2y ago
I think what it comes down to is accuracy vs speed. OpenAI clearly took steps here to improve the accuracy of the output which is critical in a lot of cases for application. Even if it will take longer, I think this is a good direction. I a
96.
by trash_cat 2y ago
It is open in the sense that everyone can use it.
97.
by trash_cat 2y ago
How do you mean quickly? It probably will take a while for it to output the final answer as it needs to re-prompt itself. It won't be as fast as 4o.
98.
by trash_cat 2y ago
I have a math disability called dyscalculia. Which basically means "number blindness". Math is about manipulating numerosities, which has nothing to do with problem solving.
99.
by trash_cat 2y ago
Now combine this with Jupyter notebooks and we are good to go. I'd pay for that.
100.
by trash_cat 2y ago
This is a strange example. A LLM would write a script for you to find the amount of "Wizard" being mentioned. That is what it always doing when it comes to numbers because it knows that it is its weakness. Edit: Going from countin
101.
by trash_cat 3y ago
Sure, I could it put it less definitively, but realistically, what else can it be? The transformer won't change much and all of the models, at the core use it. It's a closely guarded secret because it's easy to replicate.
102.
by trash_cat 3y ago
It's funny because GPT-4 is actually a pile of 3.5s. You just need to set it up correctly.
103.
by trash_cat 3y ago
Is this not an incredibly expensive/unsustainable method? I agree with the sentiment that MoE is the way to go as the newer models will probably see diminishing returns. But the compute for a single prompt will suddenly increase 7-15 f
104.
by trash_cat 3y ago
It's not cringe because of lack insight into progress, quite the opposite. It's cringe because the author does not understand what an LLM is and what it can be used for.
105.
by trash_cat 3y ago
Can you be specific what ideas of his aren't scientific? It's true that science requires replication but he deals with models but perhaps uses bad studies to support it. It's like saying he should replicate the theory of ev
106.
by trash_cat 3y ago
I did a Cog Sci bachelor and this is the conclusion I made I finished my studies. A lot of Thinking Fast and Slow are summaries of research done in the field over decades. In particular, biases and intuition as internalized knowledge/e
107.
by trash_cat 3y ago
We project everything onto AIs. Unbias in LLMs doesn't exist.
108.
by trash_cat 3y ago
I haven't tried it personally, and I'm sure I will be blown away. But the novelty will wear off fast. There is no way I would use a headset for hours.
109.
by trash_cat 3y ago
I feel like this applies as much today...
110.
by trash_cat 3y ago
Getting fitter as in working out, or losing weight...?
111.
by trash_cat 3y ago
What a refreshing and positive perspective on the matter. I couldn't agree more.
112.
by trash_cat 3y ago
This is very exciting and I think this is the future of AI until we get another, next-gen architecture beyond transformers. I don't think we will get a lot better until that happens and the effort will go into making the models a lot c
113.
by trash_cat 3y ago
You left the combatting child exploitation part out. How does pulling troops from ME will achieve that?
114.
by trash_cat 3y ago
I assume they mean "...more targeted surveillance based on reasonable suspicion".
115.
by trash_cat 3y ago
I geniounly don't understand why this comes as a surprise. If it was a realtime video with realtime responses it would have been some other generational leap beyond transformers. Not only that, it would somehow deal with video + audio
116.
by trash_cat 3y ago
You will find Veritasium's video on night vision goggles [1] very interesting. He explains well how they work, and the pros and cons of each implementation and generation. [1] https://www.youtube.com/watch?v=UAeJHAFjwPM&
117.
by trash_cat 3y ago
Here is what we have so far, correct me if I'm wrong: Ultra Release: Scheduled for early next year. Pro with Bard: Positioned between GPT-4 and Ultra in terms of performance. Currently available in the US only. Benchmarking Notes: The
118.
by trash_cat 3y ago
If I go to Bard, it specifically says that it' PaLM2 (on the side).
119.
by trash_cat 3y ago
Clearly not a meme fan. To be fair, that zoomer terminology is used on Instagram and TikTok. Pretty sure HN people are not into it.
120.
by trash_cat 3y ago
Not getting into programming because anyone can use an LLM to do programming does not suddenly fill the productivity needs in an absolute sense. People can do more with less which also means we will quickly raise the demand for the new thin

More ›