Pretraining and finetuning scripts for Amharic LLaMA adapted from:
https://github.com/ymcui/Chinese-LLaMA-Alpaca
https://github.com/facebookresearch/llama-recipes
Llama-2-Amharic weights: https://huggingface.co/iocuydi/llama-2-amharic-3784m Can be run with the inference script in this repo. Pretrained on 3.784b Amharic tokens.
Additional training, eval, translation, data processing, and LLaVA-related scripts/models to be released, along with datasets
https://medium.com/@garrilogistics/llama-2-amharic-llms-for-low-resource-languages-d6fb0ba332f4