3 ms·
Are there any "small" LM can be deployed locally that understands formats and specs and generate outputs?
by est 4y ago
Are there any "small" LM can be deployed locally that understands formats and specs and generate outputs?
- rodrigobellusci 4y agoAlpaca [1], perhaps. It's based on facebook's model (LLaMA) and its been trained on a conversational style, same as chat gpt. I don't know if it can produce code, though. [1] https://github.com/antimatter15/alpaca.cpp https://github.com/antimatter15/alpaca.cpp
- est 4y agoI mean anybody had success fine-tuning it? How hard could it be and how long does it take?
- rodrigobellusci 4y agoYeah, anybody can clone Alpaca and fine tune it. From what I can tell setting things up to train the model is easy. Then you have to generate the training data (can be done using GPT3's API), fine tune Alpaca, and then evaluate it. Haven't done it myself but I believe you can find more info here [1] [1] https://crfm.stanford.edu/2023/03/13/alpaca.html https://crfm.stanford.edu/2023/03/13/alpaca.html