Promote and proliferate local LLMs. If you use GPT, you're giving OpenAI money to lobby the government so they'll have no competitors, ultimately screwing yourself, your wallet, and the rest of us too. OpenAI has no moat, unless you give them money to write legislation. I can currently run some scary smart and fast LLMs on a 5 year old laptop with no GPU. The future is, at least, interesting.
There's no need to run locally if you aren't utilizing 8 hrs/day.
You can rent time on a hosted GPU, sharing a hosted model with others.
A reminder that llama isn't legal for the vast majority of use cases. Unless you signed their contract and then you can use it only for research purposes.
We don’t actually know that it’s not legal. The copyrightability of model weights is an open legal question right now afaik.
It doesn't have to be copyrightable to be intellectual property.
Biggest news here from a capabilities POV is actually the gpt-3.5-turbo-instruct model. gpt-3.5-turbo is the model behind ChatGPT. It's chat-fine-tuned which makes it very hard to use for use-cases where you really just want it to obey/complete without any "chatty" verbiage. The "davinci-003" model was the last instruction tuned model, but is 10x more expensive than gpt-3.5-turbo, so it makes economical sense to hack…
What's the difference between chat and instruction tuning?
Right now the closed models are incredibly higher quality than the open models. They're useful as a stopgap for 1-2 years in hopes/expectation of open models reaching a point where they can be swapped in. It burns cash now, but in exchange you can grab more market share sooner while you're stuck using the expensive but high quality OpenAI models. It's not cost-effective, but it may be part of a valid business plan.
If you're finetuning your own model, the closed models being "incredibly higher quality" is probably less relevant.
Finetuning a better model still yields better results than finetuning a worse model.
They're deprecating all the completion/edit models. The chat models constantly argue with you on certain tasks and are highly opinionated. A completion API was a lot more flexible and "vanilla" about a wide variety of tasks, you could start a thought, or a task, and truly have it complete it. The chat API doesn't complete, it responds (I mean of course internally it completes, but completes a response, rather than a…
Unfortunately their decisions are driven by model usage: gpt-3.5-turbo is the most used one (probably due to the low price and similar result)
Promote and proliferate local LLMs. If you use GPT, you're giving OpenAI money to lobby the government so they'll have no competitors, ultimately screwing yourself, your wallet, and the rest of us too. OpenAI has no moat, unless you give them money to write legislation. I can currently run some scary smart and fast LLMs on a 5 year old laptop with no GPU. The future is, at least, interesting.
Care to share some links? My lack of GPU is the main blocker for me from playing with local-only options. I have an old laptop with 16GB RAM and no GPU. Can I run these models?
Biggest news here from a capabilities POV is actually the gpt-3.5-turbo-instruct model. gpt-3.5-turbo is the model behind ChatGPT. It's chat-fine-tuned which makes it very hard to use for use-cases where you really just want it to obey/complete without any "chatty" verbiage. The "davinci-003" model was the last instruction tuned model, but is 10x more expensive than gpt-3.5-turbo, so it makes economical sense to hack…
What's the difference between chat and instruction tuning?
no expert, but from my messing around I gather the chat models are tuned for conversation, for example, if you just say 'Hi', it will spit out some 'witty' reply and invite you to respond, it's creative with it's responses. On the other hand, if you say 'Hi' to an instruct model, it might say something like, I need more information to complete the task. Instruct models are looking for something like 'Write me a twitter bot to make millions'... in this case, if you ask the same thing again, you are somewhat more like to get the same, or similar result, this does not appear so true with a chat model, perhaps a real expert could chime in :)
With the latest 3.5-turbo, you can try forcing it to call your function with a well-defined schema for arguments. If the structure is not overly complex, this should work.
It's great at returning well-formatted JSON, but it can hallucinate arguments or values to arguments.
i’ve had it come up with new function names, or prepend some prefix to the names of functions. i had to put some cleverness in on my end to run whatever function was close enough.