gpt-2-simple VS gpt_index

Tap the contact's name in WhatsApp (I think it only works on a phone) and at the bottom of that screen there's Export Chat.

For finetuning GPT-2 I think I used this thing on Google Colab. (My friend ran it on his GPU, it should be doable on most modern-ish GPUs.)

https://github.com/minimaxir/gpt-2-simple

I tried doing something with this a few months ago though and it was a bit of a hassle to get running (needed to use a specific python version for some dependencies...), I forget the details sorry!

I mentioned in a different reply that I used https://github.com/minimaxir/gpt-2-simple

You will need a few hundred bucks, python experience, and a simple implementation such as this repo https://github.com/minimaxir/gpt-2-simple

I just used the gpt-2-simple python package and ran it overnight in an jupyter notebook, but you could copy the code to any python compiler and it should also work.

I'll note that if you're just doing text generation it is a simple project as far as ML goes, there are some nice libraries you can use that require minimal ML knowlege -eg https://github.com/minimaxir/gpt-2-simple

Then, I use the GPT-2 AI libraries, wrapped in a python library GPT-2 Simple to generate the content. My actual code is basically just their code sample, so basically 6 lines of python. With GPT-2, you train the existing AI to your specific dataset, which in my case is this text file of tweets.

gpt_2_simple is your best bet! Its super easy to use, you just need to downgrade TensorFlow and some other packages in your environment.

I used the GPT-2 Simple program by minimaxir to train the algorithm on every card in Magic's history that was released in a main expansion. Then I generated about 2,000 (it was actually more, but the algorithm really liked giving me cards that already exist) new names which I searched through to find the best ones.

Might I suggest starting Starting here, to learn on Simple GPT2. They have a Google Colab Notebook if your CPU GPU is shit, and what helped me learn best is dissect the code, and basically make my own Colab notebook piece by piece, learning what each function does.

The last couple of months, I've been running a Twitter bot that posts GPT-2-generated content, trained off of Tweets from existing accounts using gpt-2-simple. In my more recent training sessions, it seems like the quality of the output has been decreasing; it often gives outputs that are just barely modified from the original training data, if not verbatim.

LLAMA Index Github repository

Zuwendungen kommen auch so langsam ( LLamaIndex ) https://github.com/jerryjliu/gpt_index

This is exactly what LlamaIndex is meant to solve!

A set of data structures to augment LLM's with your data: https://github.com/jerryjliu/gpt_index

This is what we've designed LlamaIndex for! https://github.com/jerryjliu/gpt_index. Designed to help you "index" over a large doc corpus in different ways for use with LLM prompts.

https://github.com/jerryjliu/gpt_index might be close to what you need.

I would start with https://gpt-index.readthedocs.io/en/latest/ and https://langchain.readthedocs.io/en/latest/

People have been doing this with https://github.com/jerryjliu/gpt_index

Here is a link to the repository. Take a look at the overview section of the readme. https://github.com/jerryjliu/gpt_index

(creator of gpt index / llamaindex here https://github.com/jerryjliu/gpt_index)

Funny that we had just rebranded our tool from GPT Index to LlamaIndex about a week ago to avoid potential trademark issues with OpenAI, and turns out Meta has similar ideas around LLM+llama puns :). Must mean the name is good though!

Also very excited to try plugging in the LLaMa model into LlamaIndex, will report the results.