I am really new to this field, and just have a little knowledge about AI.
Recently I am doing about training an open source large language model (LLM) with my own dataset, and make it answer the user’s question based on the dataset (If not mistaken this processing is called ‘fine-tuning’ but I am not sure).Afterward, this model will be used to develop a chatbot.
I am struggle with which model is better to use in my situation. Based on my research, I am thinking about using Mistral-7B or llama2-7B, but I don’t really sure which of them is suitable for my work, or both of them is not suitable.
Currently is still doing an experiment only so I think if can, I will run the model in colab or kaggle notebook. About the hardware requirements, I only have maximum 128 GB RAM and about 1TB Storage. The dataset is in English, but probably in the future development of chatbot need to consider about the others language such as Chinese also (multilingual capabilities).
Can someone give some advice or suggestion based on my situation. It will really help for my work