NVIDIA NIM Release

Overview of changes:

• Upgrades to OpenAIChat initialisation to allow support for usage of NVIDIA NIM models. Default nvidia NIM model is meta/llama-3.1-405b-instruct.
• Re-engineering and refinement of some prompts to get more accurate responses from language model.
• Custom timeout and retry mechanism to avoid users being on hold for unacceptably long periods of time. Default is 10s, can be configured to be higher in .initialise line for OpenAIChat in index.js

What's Changed

• Upgrades to use of NVIDIA NIM Llama Model by @Prakhar896 in #7
• Low-level OpenAI API upgrades for timeout-and-retry mechanism by @Prakhar896 in #9
• Re-engineered prompts to better suit NVIDIA NIM by @Prakhar896 in #10

Full Changelog: v1.0...v1.1

© 2024 BrainBloomAI Team. All rights reserved.

Stable Release

Overview of changes:

• Upgraded AI prompts to be more specially engineered for the GPT model's behaviour.
• Added export ability which processes, orders and synthesises the vast amounts of data points into real-time generated CSV, TXT and JSON formats for analysis.
• Debug endpoints for supers.
• First stable release of BrainBloomAI Express API. Version 1.0 🎉 !

New Contributors:

• @sarahzsu made their first contribution in #5

Full Changelog: v0.1...v1.0

© 2024 BrainBloomAI Team. All rights reserved.

Stable Beta Release

Before upgrades to AI system.

Full Changelog: https://github.com/BrainBloomAI/Backend/commits/v0.1