Open-source embeddings model from Microsoft initialized from Mistral-7B-v0.1, achieving state-of-the-art BEIR score of 56.9 for English text embedding and retrieval tasks with 4096-dimensional vectors.
E5-Mistral is an open-source embeddings model developed by Microsoft, released under the MIT license. This E5 embedding model by Microsoft is initialized from Mistral-7B-v0.1 and fine-tuned on a mixture of multilingual datasets.
The task definition should be a one-sentence instruction that describes the task. This is a way to customize text embeddings for different scenarios through natural language instructions.
Built with PyTorch, it generates high-quality vector representations useful for:
Since Mistral-7B-v0.1 is mainly trained on English data, it's recommended to use this model for English only.
For e5-mistral-7b-instruct, it would require transformers>=4.34 to load Mistral model.
The model is available on:
Free and open-source under MIT license.
Loading more......