Overview

VMware/open-llama-13b-open-instruct is a 13 billion parameter instruction-tuned language model developed by VMware. It is built upon the Open Llama architecture and has been fine-tuned using a proprietary dataset, open-instruct-v1, which integrates data from OASST, Dolly, and HHRlhf. This model is explicitly licensed for commercial use, making it a viable option for various business applications.

Key Capabilities

• Instruction Following: Optimized to understand and execute instructions provided in the Alpaca prompt template format.
• Commercial Viability: Licensed under terms that permit commercial deployment.
• Open Llama Base: Benefits from the foundational capabilities of the Open Llama 13B model.

Important Considerations

• Tokenizer Usage: Users must set use_fast = False when instantiating the tokenizer to avoid incorrect encoding issues.
• Code Performance: The model may exhibit limitations in code-related tasks due to the tokenizer's handling of multiple spaces.

Good for

• Applications requiring a commercially usable, instruction-tuned LLM.
• General-purpose instruction-following tasks where the Alpaca prompt format is suitable.