Improving Text Embeddings with Large Language Models: Analysis of Training Hyperparameters | HackerNoonMistral-7B outperform LLaMA-2 7B in model initialization, highlighting effective performance in multilingual tasks.