jwu323/origin-llama-7b

TEXT GENERATIONConcurrency Cost:1Model Size:7BQuant:FP8Ctx Length:4kArchitecture:Transformer Cold

The jwu323/origin-llama-7b model provides the original 7 billion parameter LLaMA weights, designed for research and development purposes. This foundational model has a context length of 4096 tokens and is intended for users who have been granted access to the LLaMA weights under its non-commercial license. It serves as a direct source for the initial LLaMA architecture, facilitating conversion to the Hugging Face Transformers format.

Loading preview...

jwu323/origin-llama-7b: Original LLaMA Weights

This repository hosts the original 7 billion parameter LLaMA model weights. It is specifically intended for users who have already obtained access to the LLaMA model under its non-commercial license but require the original weights, potentially for conversion to the Hugging Face Transformers format or if their initial copy was lost.

Key Characteristics

  • Model Family: LLaMA
  • Parameter Count: 7 billion parameters
  • Context Length: 4096 tokens
  • License: Non-commercial (requires prior access grant)

Intended Use

This model is primarily for:

  • Researchers and developers with existing LLaMA access.
  • Facilitating conversion of original LLaMA weights to the Hugging Face Transformers format.
  • Ensuring access to the foundational LLaMA-7b weights for licensed users.