Tried to put it into my svelte project. Build failed. Went onto github to report it. A reproduction url is required ... so not submitting a bug ... and looking for something else to use ...
Retrieval Augmented Generation. Fancy way of saying, retrieve chunks from your document corpus similar to your input using a similarity (mostly cosine) of embedding vectors of the chunks and input vectors, then inject those relevant chunks into your prompt to the LLM. Useful for Document Intelligence.
Retrieval Augmented Generation - using search (usually with some kind of semantic component) to find relevant context and provide it to the language model to help it respond, give it knowledge about a specific document, etc.
This is how I felt when trying to learn Elm: the program had to be correct, exactly correct, or it wouldn't work. You had to make every piece, every function, fit precisely, to define it's shape, it's exact inputs and outputs and effects ... in the end I found it very restrictive. I like the idea of loose-ness by default and adding contraints gradually (like javascript -> typescript).
Someone on YouTube mentioned that LK99 is ceramic ie not metal which nullifies many of its potential uses even if it does turn out to be superconducting
That's how they are trained initially, but the resulting model isn't all that useful (was SOTA two years ago but this field moves fast).
A lot of the utility comes from the later finetuning. You can see this using the examples from the article, every mistake they identify with GPT-3 (which is the unfinetuned version) is answered correctly by chatGPT, which has gone through an extensive finetuning process called RLHF.
That's how the text decoder works, but the model gets to define "most likely" and an RLHF model uses this to make the text decoder produce useful answers instead.
If you don't care what exact tool in particular, https://github.com/AUTOMATIC1111/stable-diffusion-webui is the easiest to install I think and gives lots of toys to play with (txt2img, img2img, teach it your likeness, etc.)
If you're used to installing python packages it should be relatively easy. There are other projects with nice UIs but that's not what this library is for.
