Mxbai Embed Large 335M - Details

The Mxbai Embed Large 335M is a large language model developed by Mixedbread, featuring 335m parameters under the Apache License 2.0. It is designed to achieve state-of-the-art performance on MTEB while ensuring strong generalization capabilities.

The mxbai-embed-large-v1 is a sentence embedding model developed by Mixedbread as part of the crispy sentence embedding family. It supports retrieval tasks, Matryoshka Representation Learning, and binary quantization. The model is multilingual and multimodal, designed for efficient sentence embeddings with applications in semantic search, clustering, and document retrieval. Its architecture emphasizes scalability and performance across diverse language and modalities.

The Mxbai Embed Large 335M model features 335m parameters, placing it in the small category of open-source LLMs, which are fast and resource-efficient for simple tasks. Its 4k context length falls under short contexts, making it suitable for short tasks but limited in handling long texts. These specifications balance accessibility and performance, ideal for applications requiring moderate computational resources.

• Name: Mxbai Embed Large 335M
• Parameter Size: 335m
• Context Length: 4k

The Mxbai Embed Large 335M model is designed for tasks that require efficient and effective sentence embeddings, with possible applications in areas like retrieval, semantic search, and clustering. Its 335m parameter size and 4k context length suggest it could be used for possible tasks involving text analysis, information organization, or data categorization. However, these possible uses would need thorough testing to ensure compatibility with specific workflows or datasets. The model’s multilingual and multimodal capabilities further expand its possible utility in scenarios requiring cross-lingual or cross-modal understanding. While the intended uses include retrieval, semantic search, and clustering, the possible applications may vary depending on the context and requirements of the user.

• retrieval
• semantic search
• clustering

The Mxbai Embed Large 335M model could be used for possible applications such as retrieval tasks, semantic search, clustering, and cross-lingual information organization. Its possible suitability for these areas stems from its multilingual and multimodal design, which might enable possible improvements in organizing large text datasets or enhancing search efficiency. However, these possible uses would require thorough evaluation to ensure alignment with specific requirements. The possible effectiveness of the model in tasks like document clustering or semantic similarity ranking remains to be validated through testing. Each possible application must be carefully assessed before deployment to confirm its viability.

• retrieval
• semantic search
• clustering
• cross-lingual tasks

The Mxbai Embed Large 335M model’s medium q4 version requires hardware capable of handling 335m parameters with quantized precision, likely needing a GPU with at least 8GB VRAM for efficient operation. This version balances precision and performance, making it suitable for systems with moderate computational resources. However, possible variations in implementation may affect these requirements. The available quantized versions include fp16, which is optimized for general-purpose use.

• fp16

The Mxbai Embed Large 335M is a large language model developed by Mixedbread, featuring 335m parameters under the Apache License 2.0, designed for state-of-the-art performance on MTEB with strong generalization. It is part of the crispy sentence embedding family, supporting retrieval tasks, Matryoshka Representation Learning, and binary quantization, while being multilingual, multimodal, and suitable for semantic search, clustering, and document retrieval.

Huggingface Model Page
Ollama Model Page