georgesung/llama2_7b_openorca_35k
TEXT GENERATIONConcurrency Cost:1Model Size:7BQuant:FP8Ctx Length:4kLicense:otherArchitecture:Transformer0.0K Cold
georgesung/llama2_7b_openorca_35k is a 7 billion parameter Llama-2 model fine-tuned by georgesung. It was trained using QLoRA on a 35k subset of the OpenOrca dataset, optimizing it for instruction-following and helpful AI assistant tasks. This model is designed for general-purpose conversational AI applications, leveraging its fine-tuning on diverse instruction data.
Loading preview...
Overview
This model, georgesung/llama2_7b_openorca_35k, is a 7 billion parameter Llama-2 variant that has been fine-tuned by georgesung. It leverages the efficient QLoRA method for training, making it accessible for fine-tuning on more modest hardware (e.g., a 24GB GPU).
Key Capabilities
- Instruction Following: Fine-tuned on a 35k subset of the OpenOrca dataset, enhancing its ability to understand and respond to instructions.
- Conversational AI: Trained with a specific prompt style (
### System:,### Instruction:,### Response:) that aligns it well with helpful AI assistant roles. - Efficient Fine-tuning: The use of QLoRA allows for effective fine-tuning of large models with reduced memory requirements.
Good For
- General-purpose AI assistants: Its instruction-tuned nature makes it suitable for various conversational tasks.
- Prototyping and development: The model's size and fine-tuning method offer a good balance of performance and resource efficiency for developers.
- Further experimentation: The training code is publicly available here, enabling users to reproduce or extend the fine-tuning process.