peterthehacker

Congrats! I agree that this is a very important problem to solve. Can you elaborate on some of the specifics of how LTSE will incentive investors to think long-term?

I’m imagining some strict trading frequency limits. Like an inventor can only trade a specific stock once a year. Do I have the right idea?

It’s not fair to critique a film or video game based on its trailer. The phrase “don’t judge a book by its cover” comes to mind.

Let’s suspend judgment of Cyberpunk 2077 until it’s released in a couple months. They surprised everyone with how awesome Witcher 3 was, so let’s give CD Projekt the benefit of the doubt.

Bitcoin has a large (100M+) user base (or at least it seems like it) but I can’t think of another blockchain app that has reached that level. Ethereum is more like a framework or platform as I understand it so I don’t think it counts in the same way.

Besides bitcoin are there any other blockchain apps that have large user bases?

This is a great point!

Reading this paper reminds me how important it is for AI to continue to evolve it’s core algorithms.

Deep learning models are effectively pattern matching machines that cannot separate causality from correlation. As we throw more compute and $$$ at deep learning models we will experience diminishing returns in performance because of this.

For us to achieve AGI[0], the holy grail of AI, we will need to develop algorithms that can recognize causality somehow. Judea Pearl’s “The Book of Why”[1] does a great job articulating why this is important. Deep learning is a big leap forward and we’re only beginning to see its impacts, but it’s not sufficient to achieve AI’s most ambitious goals.

[0] https://en.m.wikipedia.org/wiki/Artificial_general_intellige...

[1] https://www.amazon.com/Book-Why-Science-Cause-Effect/dp/0465...

Aren’t companies like google[0] and nvidia[1] already doing this?

The paper’s point is that eventually we will reach computing power limits and then we will have to improve the deep learning algorithm’s efficiency to continue to improve. From the abstract:

> Continued progress in these applications will require dramatically more computationally-efficient methods, which will either have to come from changes to deep learning or from moving to other machine learning methods.

[0] https://cloud.google.com/tpu

[1] https://www.nvidia.com/en-us/data-center/v100/

Thanks for sharing!

Can you elaborate on “What’s next”?

> User callbacks- so that application shows not just pre-calculated visualizations but also can fetch data from a store and process it in real-time.

How are you envisioning this working? Will dstack be like a database? How will “user callbacks” be triggered?

Are deepfake face swaps a real problem yet? I can’t recall any major controversies in recent history that were caused by a deepfake face swap.

> This technique can be used to create compromising videos of virtually anyone, including celebrities, politicians, and corporate public figures.

I’ve read a lot of concerned comments like this but I haven’t read about any real world examples of a controversy caused by a deepfake.

That’s what it looks like. There’s a comment in this issue calling that out also. [0]

Considering how inaccessible it is I’m curious why there’s so much hype around it. Just the cost/benefit of cloud compute resources needed to run the model looks really bad. All of these hypothetical use cases being discussed in other threads would be financially unsustainable.

Granted, it’s still yet to be seen how the smaller models will perform. Looking forward to experimenting with those once they’re released!

[0] https://github.com/openai/gpt-3/issues/1#issuecomment-636008...

Like many others in HN, I’m interested in learning more about GPT-3 and experimenting with the model myself but I was surprised to find that it’s not openly available and that the repo was archived, perhaps for the reasons discussed in this issue.

Maybe someone here on HN could help shed light on what’s going on and clarify if/when a GPT-3 model will be available?

I think that it’s the opposite. This algorithm requires many examples of text on the specific topic. Probably more than most humans would require.

> While typically task-agnostic in architecture, this method still requires task-specific fine-tuning datasets of thousands or tens of thousands of examples [0]

I don’t know what constitutes an example in this case but let’s assume it means 1 blog article. I don’t know many humans that read thousands or tens of thousands of blog articles on a specific topic. And if I did I’d expect that human to write a much more interesting article.

To me, this and other similar generated texts from OpenAI feel bland / generic.

Take a listen to the generated music from OpenAI - https://openai.com/blog/jukebox/. It’s pretty bad, but in a weird way. It’s technically correct - in key, on beat, ect. And even some of the music it generates is technically hard to do, but it sounds so painfully generic.

> All the impressive achievements of deep learning amount to just curve fitting Judea Perl [1]

This comment was written by a human :)

[0]https://arxiv.org/abs/2005.14165 [1]https://www.quantamagazine.org/to-build-truly-intelligent-ma...

I'm trying to figure out how to run these benchmarks on my own machine and experiment with some tweaks to the implementation, but it's unclear how to run these benchmarks from start to finish. I don't see any instructions for running the benchmarks in the github repository.

@calpaterson can you provide guidance?

I'd like to try an alternative query pattern. The current pattern implemented in the benchmarks is select 1 row in 1 query. I'd like to try an implementation with 2 queries - select count() from table and select * from table limit 10, which is a very common pattern for a REST list view. I would hypothesize that the async apps would perform better in this case, but I'm curious what this benchmark with show.