search

ngruver / llmtime

https://arxiv.org/abs/2310.07820
MIT License
628 stars 139 forks source link

Added latest gpt models, mistral local and mistral api #23

Closed zaizou closed 5 months ago

zaizou commented 5 months ago

In this updated version, I integrated GPT Turbo by making some modifications to the models/gpt.py file. Additionally, I incorporated two methods for utilizing Mistral to fully explore its capabilities: The local model and the Mistral API. The "mistralai" Python package has been included in the install.sh script. Within the Mistral API, users have the option to choose from three models: tiny, small, and medium. The NLL for Mistral 7b in the local setup was obtained from llama, while it has not yet been implemented for the Mistral API. In future iterations, the Mistral embedding API can be utilized for tokenization. Although the current version utilizes the existing tokenization method, an update using the Mistral embedding API may be implemented soon.

zaizou commented 5 months ago

any question please contact me at : medjdoub.bouzid.ini@gmail.com

ngruver commented 5 months ago

LGTM. Thanks for the contribution!