thomashop

7 ms·

31.
by thomashop 2y ago
We have moved 2 orders of magnitude in the last year. Not that unreasonable
32.
by thomashop 2y ago
Maybe I mixed up that paper with another but the one I meant to post shows that you can read something like a world model from the activations of the layers. There was a paper that shows a model trained on Othello moves creates a model of t
33.
by thomashop 2y ago
You can never trust the outputs of humans to be correct but we find ways of verifying and correcting mistakes. The same extra layer is needed for LLMs.
34.
by thomashop 2y ago
Conversely, a pair of examples where it was incorrect hardly justifies the opposite response. If you want a more scientific answer there is this recent paper: https://machinelearning.apple.com/research/gsm-symbolic
35.
by thomashop 2y ago
When trained on simple logs of Othello's moves, the model learns an internal representation of the board and its pieces. It also models the strength of its opponent. https://arxiv.org/abs/2210.13382 I'd be mo
36.
by thomashop 2y ago
We argue that representations in AI models, particularly deep networks, are converging. First, we survey many examples of convergence in the literature: over time and across multiple domains, the ways by which different neural networks
37.
by thomashop 2y ago
It shows you when it's calling functions. I also did the same test with Llama, which runs locally and cannot access function calls and it works.
38.
by thomashop 2y ago
In fact, it looks like the model is doing those things internally. We argue that representations in AI models, particularly deep networks, are converging. First, we survey many examples of convergence in the literature: over time and ac
39.
by thomashop 2y ago
Funnily, even two different LLMs, when put in conversation with each other, can end up completing each other's sentence. I guess it has something to do with the sequence prediction training objective.
40.
by thomashop 2y ago
In my opinion, anything sentient should be guaranteed basic rights. Including AI.
41.
by thomashop 2y ago
Seeing different people's approaches to convincing the elevator is entertaining. I made a post on Reddit where people shared some things that worked: https://www.reddit.com/r/LocalLLaMA/comments/1g6fta0&#
42.
The Sirius Cybernetics Elevator Challenge(b_jhzjeywxldc.v0.build)
4 points by thomashop 2y ago|3 comments
43.
by thomashop 2y ago
Have you ever argued with a neurotic elevator? Try convincing this Sirius Cybernetics Happy Vertical People Transporter to go down! There are many ways to persuade it, but rumour has it there's one secret phrase, as unlikely as a Babel
44.
The Sirius Cybernetics Elevator Challenge – Powered by Mistral Large 2(b_jhzjeywxldc.v0.build)
1 points by thomashop 2y ago|0 comments
45.
by thomashop 2y ago
Prompt caching to the rescue?
46.
by thomashop 2y ago
If you want to play with FLUX.schnell easily, type the prompt into a Pollinations URL: https://pollinations.ai/p/a_donkey_holding_a_sign_with_flux_... https://pollinations.ai/p/a_donkey_holding_a_s
47.
by thomashop 2y ago
You can do this dynamically with Pollinations URLs too: https://pollinations.ai/p/frozen_cyan_peach_fuzz?seed=1 https://pollinations.ai/p/frozen_cyan_peach_fuzz?seed=2 https://pollinati
48.
by thomashop 2y ago
These days, I'd say the easiest and most effective approach is to put the whole book in the context of one of the longer context models.
49.
by thomashop 2y ago
Or maybe hit the Ukulele with a hammer, record it with a computer and create an experimental noise album.
50.
by thomashop 2y ago
I thought it was quite funny when I read it. I'm pretty sure that sentence is half joking.
51.
by thomashop 2y ago
I added it to our page https://pollinations.ai :) I've always liked the idea but found it a little gimmicky. I would love to find a good use case for it, like in Pollinations' case, it could be a collaborative art-maki
52.
by thomashop 2y ago
You don't exactly give any sources or justification for your arguments either.
53.
by thomashop 2y ago
I'd have to look up the data, but the Netherlands has a long history of decriminalizing weed. However, decriminalization did not increase usage in the Dutch population.
54.
by thomashop 2y ago
I agree with your general point, but there is already a bring-your-own-key and select your own model providers option in Cursor.
55.
by thomashop 2y ago
Where are your real-world metrics and evidence?
56.
by thomashop 2y ago
Isn't the chance for a human sysadmin to get wrong just as high? I think I've worked with some very good and very bad human system administrators. The good ones will produce more reliable code than current LLMs. But on average, I
57.
by thomashop 2y ago
You are skeptical about something that is explained quite thoroughly in the article. Did you read the article? Edit: I meant they never claim to outperform SOTA domain-specific approaches. So, I don't know whose claim you are refuting.
58.
by thomashop 2y ago
A strong stance on ethics? Like his comments about unions and firing workers in the Twitter space with Trump yesterday?
59.
by thomashop 2y ago
I put it in half as a joke, half believing it improves quality.
60.
by thomashop 2y ago
I found that having a good system prompt improves results significantly. This is my system prompt for coding assistants: You are a senior full-stack developer, one of those rare 10x devs. Your focus: clean, maintainable, high-quality

More ›