lgas

19 ms·

271.
by lgas 3y ago
You're confusing simple and easy.
272.
by lgas 3y ago
How do you do it on Pixel phones?
273.
by lgas 4y ago
Regulation can't ensure anything, it can only change incentives.
274.
by lgas 4y ago
Actually they see tokens, which are, usually, parts of words.
275.
by lgas 4y ago
You wouldn't necessarily expect prompts compressed with one model to work with another though, right? Presumably what's going on is tied to the specified vector space, right?
276.
by lgas 4y ago
If the goal is just to have the model break down each task into sub tasks until they are small enough to perform, why not implement the recursion in the code that calls the models where it's a solved problem? Even if you got this work
277.
by lgas 4y ago
> Just don't hire a junior because you are afraid of hiring someone senior as your first hire. I'd go so far as to say don't hire a junior person as your first hire. Someone senior can help you with hiring in general and
278.
by lgas 4y ago
I believe that's pretty close to what https://bittensor.com/ does.
279.
by lgas 4y ago
What's the actual goal here? If you got it working really well, what is it that would you be able to do with it better than using some other approach? As to getting the math/logic working better in the prompt, it seems like the o
280.
by lgas 4y ago
Maybe they meant the gallery itself is minimal :)
281.
by lgas 4y ago
Hi, since you're commenting here and appear to be interested in exploring things further, I just wanted to point out that once GPT-4 is available through the API you could use a LangChain Agent[1] to maintain the board state externally
282.
by lgas 4y ago
Not the performance, but I thought this was illuminating: https://i.imgur.com/zBAkjl1.jpg
283.
by lgas 4y ago
For the most part, they still do.
284.
by lgas 4y ago
The models embedded in Bing and ChatGPT have limited agency but through the APIs people have given them all sorts of agency (eg using https://langchain.readthedocs.io/en/latest/modules/agents/to... ). I&
285.
by lgas 4y ago
This worked for me with GPT-4: Let's play tic tac toe. After every move please check the game state to see if anyone has won. You go first.
286.
by lgas 4y ago
Are you suggesting that the content they have consumed is not part of their experience?
287.
by lgas 4y ago
It is, if you request and get approved. https://github.com/facebookresearch/llama
288.
by lgas 4y ago
That's the point, even C64 tier examples in one shot or few shot learning do wonders for changing the behavior of the model.
289.
by lgas 4y ago
So, are you working on algorithms and OOP then?
290.
by lgas 4y ago
How about https://arxiv.org/abs/1803.05316 ?
291.
by lgas 4y ago
I'd suggest providing some justification for your declaration if you want anyone to listen to you.
292.
by lgas 4y ago
I haven't tried this with recent versions, but this was great fun early on: https://github.com/f/awesome-chatgpt-prompts#act-as-an-ai-tr...
293.
by lgas 4y ago
The project (Ogma) looks like it involves a lot of parsing, converting between data types, and the emitting new code. Haskell is a pretty good fit for that type of domain.
294.
by lgas 4y ago
I'm pretty sure this is built on the OpenAI API, so not so much a competitor as a new sales channel.
295.
by lgas 4y ago
If you google, you will find many, eg. https://www.chatgptalexa.com/
296.
by lgas 4y ago
Generally not a good idea to have these at all.
297.
by lgas 4y ago
I don't know... LLMs are just functions from text to text. Seems like the same skillset is still valuable. You just have more amazing functions at your disposal.
298.
by lgas 4y ago
Why bother writing each backend separately? https://github.com/TheAppleTucker/backend-GPT
299.
by lgas 4y ago
> IMO having a kid is the cutoff between old young and young old. As a 47 year old, I'm glad to know I will be forever young.
300.
by lgas 4y ago
> Think of ingesting a whole novel that takes place over a few years. I did exactly that with Asimov's Let's Get Together using https://github.com/jerryjliu/gpt_index . It's a short story that's

More ›