TheBloke/guanaco-7B-HF

TEXT GENERATIONConcurrency Cost:1Model Size:7BQuant:FP8Ctx Length:4kPublished:May 25, 2023License:otherArchitecture:Transformer0.0K Cold

TheBloke/guanaco-7B-HF is a 7 billion parameter language model, a re-packaging by TheBloke of Tim Dettmers' Guanaco 7B. This model is provided in fp16 Hugging Face format, resulting from merging the LoRA fine-tuning into the base model. It offers a standard 4096 token context length and is suitable for general-purpose language generation and understanding tasks.

Loading preview...

Overview

The TheBloke/guanaco-7B-HF model is a 7 billion parameter language model, specifically the fp16 Hugging Face version of Tim Dettmers' Guanaco 7B. This release by TheBloke provides the model with the LoRA fine-tuning merged into the base model, making it ready for direct use in its full precision format.

Key Characteristics

  • Model Size: 7 billion parameters, offering a balance between performance and computational requirements.
  • Format: Provided in fp16 (half-precision floating point) Hugging Face format, suitable for GPU inference.
  • Origin: Based on Tim Dettmers' Guanaco 7B, a model known for its instruction-following capabilities.
  • Context Length: Supports a standard context window of 4096 tokens.

Usage Considerations

This model is ideal for users who require a full-precision version of Guanaco 7B for various natural language processing tasks. It serves as a foundational model for further experimentation or deployment where fp16 precision is preferred. Other quantized versions (GPTQ, GGML) are available from TheBloke for different hardware and performance needs.