Title

Add Mistral-7B-Instruct-v0.2 to KAITO supported model list.

Glossary

N/A

Model description: Launched last September, Mistral-7B has quickly become a standout among large language models (LLMs) with its 7.3B parameters. It surpasses models nearly twice its size in benchmarks, showcasing its efficiency and cost-effectiveness. Mistral has also forged strategic alliances with major cloud platforms including Azure, AWS, and GCP. For more information, refer to the Mistral-7B Documentation and access the model on Hugging Face. The newly introduced Mistral-7B-Instruct-v0.1 is an instruct fine-tuned iteration of the original, optimized for understanding and executing specific instructions. This enhances mistral's utility in conversational applications.
Model usage statistics: In the past month, Mistral-7B-Instruct-v0.1 has garnered 424,580 downloads on Hugging Face, reflecting its widespread popularity. Google Trends data shows a high level of search interest in "mistral ai 7b", indicating strong market curiosity.
Model license: Mistral-7B-Instruct-v0.1 is distributed under the Apache 2.0 license, ensuring broad usability and modification rights.

The following table describes the basic model characteristics and the resource requirements of running it.

Field	Notes
Family name	Mistral
Type	`text generation`
Download site	https://huggingface.co/mistralai/Mistral-7B-Instruct-v0.2
Version	b70aa86578567ba3301b21c8a27bea4e8f6d6d61
Storage size	50GB
GPU count	1
Total GPU memory	16GB
Per GPU memory	`N/A`

This section describes how to configure the runtime framework to support the inference calls.

Options	Notes
Runtime	Huggingface Transformer
Distributed Inference	False
Custom configurations	Precision: BF16. Can run on one machine with total of 16GB of GPU Memory.