141 9 months ago

Continue finetuning https://huggingface.co/meta-llama/Llama-3.2-3B on highly curated 1.5B tokens Malaysian instruction dataset.

13 models