brucethemoose/Capybara-Tess-Yi-34B-200K

TEXT GENERATIONConcurrency Cost:2Model Size:34BQuant:FP8Ctx Length:32kPublished:Nov 18, 2023License:yi-licenseArchitecture:Transformer0.0K Cold

The brucethemoose/Capybara-Tess-Yi-34B-200K is a 34 billion parameter language model, a merge of NousResearch/Nous-Capybara-34B and migtissera/Tess-M-Creative-v1.0, based on the Yi-34B-200K architecture. This model is designed for extended context applications, supporting a 200K token context length. It is particularly suited for creative writing and complex conversational tasks, leveraging the strengths of its merged components.

Loading preview...

Model Overview

The brucethemoose/Capybara-Tess-Yi-34B-200K is a 34 billion parameter language model created by brucethemoose. It is a merge of three distinct models: NousResearch/Nous-Capybara-34B, migtissera/Tess-M-Creative-v1.0, and larryvrh/Yi-34B-200K-Llamafied, using the ties merging method. The base model for this merge is the Yi-34B-200K, known for its substantial context window.

Key Capabilities

  • Extended Context: Supports an impressive 200,000 token context length, enabling deep and long-form interactions.
  • Merged Strengths: Combines the capabilities of Nous-Capybara-34B and Tess-M-Creative-v1.0, suggesting a blend of reasoning and creative writing prowess.
  • Vicuna Syntax: Utilizes the Vicuna prompt format, making it compatible with common instruction-tuned model interfaces.

Prompt Format

The model expects prompts in the following Vicuna-style format:

SYSTEM: ...
USER: ...
ASSISTANT: ...

Users should be aware that the model may sometimes output </s> as a stop token, which might need to be added as an explicit stopping condition in generation settings.

Good For

  • Applications requiring very long context understanding and generation.
  • Creative writing tasks, storytelling, and role-playing scenarios.
  • Complex conversational agents that benefit from merged model capabilities.