A highly async actor model is something I’ve wanted to explore, and combined with a highly multi core architecture but clocked very very low, it seems like it could be power efficient too.
I was considering using go + channels for this
Ask me how I know.
iPadOS 15 hasn't been receiving the security updates for vulnerabilities that have been released for iPadOS 18.
I think you’re overthinking this. System 1 thinking as the term is being used by AI researchers means making a fast decision based on reasoning processes that are wired into your brain by evolution. For any task that humans have faced for millions of years this works well. It can also work well for experts in a domain who have practiced a task so many times that their brains have adapted to perform it unconsciously.
System 2 thinking is consciously using explicit reasoning techniques to think through a problem, slowly and rigorously, often in ways that feel unnatural due to our cognitive biases but can solve problems that System 1 is unable to.
The analogy to LLMs is straightforward: LLMs learn to solve many kinds of complex problems during training and encode processes for those specific problems. They can then perform these tasks in a single forward pass through their weights. This is System 1 for LLMs and again, works well for any task that they were exposed to repeatedly during training.
However they don’t generalize to tasks that were not well represented in the training data. Training them to use explicit reasoning strategies instead (System 2) is shown to improve performance and let them solve a broader range of problems.
e.g. currently, it's doing:
[input] -> [3 hashes] -> [single storage]
but, I'm wondering if it's better to do: [input]
-> [hash a] -> [storage a]
-> [hash b] -> [storage b]
-> [hash c] -> [storage c]
...this way, the output of the 3 hashes don't affect each other during the set and membership-check. I wonder how that would affect how much more data you can store. I'm sure someone has considered this.Relatedly, beware of the hidden ads in social media. It takes much less money to astroturf a community like reddit than to advertise through conventional means (e.g. commercials, product placements, banner ads). You may be aware of ads on youtube, but you may not be aware of the ads in the comment section. I've seen youtube comments as follows:
Ed: Christ! My stock portfolio has gone down so much since the pandemic... it hurts.
Bob: Mine did too, however in the past year I've been able to recover financially.
Ed: Really? Please help me, how did you recover your losses?
Bob: I'd hate to share my secret, but Jonathon Harris provided me with amazing investment tips.
Sal: Mr. Jonathon Harris helped 10X our portfolio! You can reach him at jonharris@gmail.com
</ad scam>That was my re-enactment of their discussion. Sure, that was an example of an obvious ad. But think of all the times you read a conversation about a product or service and became convinced to try something just based on a conversation between two supposed strangers online? I've never bought anything from a conventional ad, but I've purchased plenty of things after reading what the socials say about a product, and that makes me ashamed of myself. Word of mouth is effective for a reason, but it's insidious online. And that's where they get you. Steel your mind, when online.