Release Note: Version 0.1.5 (2023-10-19)

New Features

Auto device selection: Now supports CPU, GPU, and multi-GPU configurations.
Enhanced LoRa/QLoRa Support: Added 8-bit and 4-bit support for LoRa and QLoRa devices.
Bitsandbytes Integration: Added support for bitsandbytes data format.
Hardware-Agnostic Quantized Models: Implemented GGML for hardware-agnostic quantized models.
RAG Generation with Pinecone: Added support for RAG generation using Pinecone.
ChatGPT Finetuning: Enabled finetuning capabilities for ChatGPT models.

Fixes

ChatGPT Finetune Model Update: Updated ChatGPT finetune model to gpt-3.5-turbo-0613.

Other Changes

Refactored Completions: Code completions have been refactored for better performance.
Class Name Standardization: Corrected class names to adhere to CamelCase conventions.
Cleanup: Removed duplicated example files and ensured all necessary requirements are included.