shuttleai/shuttle-3

Hugging Face
TEXT GENERATIONConcurrency Cost:4Model Size:72.7BQuant:FP8Ctx Length:32kPublished:Oct 9, 2024License:qwenArchitecture:Transformer0.0K Warm

Shuttle-3 is a 72 billion parameter multilingual language model developed by ShuttleAI Inc., fine-tuned from Qwen-2.5-72B-Instruct. It is specifically optimized to emulate the writing style of Claude 3 models and excels in complex chat, reasoning, and agent tasks, particularly with extensive training on role-playing data. This model is designed for advanced conversational AI applications requiring nuanced prose and robust multilingual communication.

Loading preview...

Shuttle-3: Next-Generation Multilingual Chat Model

Shuttle-3, developed by ShuttleAI Inc., is a 72 billion parameter language model fine-tuned from the Qwen-2.5-72B-Instruct base. Released in beta on October 25, 2024, this model is engineered for advanced conversational AI, multilingual communication, and complex reasoning tasks.

Key Capabilities

  • Claude 3 Prose Emulation: Fine-tuned to replicate the sophisticated writing style and prose quality of Claude 3 models.
  • Role-Playing Proficiency: Extensively trained on diverse role-playing data, making it highly effective for interactive and character-driven applications.
  • Multilingual Support: Built upon a base model pretrained with a significant proportion of multilingual data, ensuring robust performance across various languages.
  • Agentic Reasoning: Designed to excel in agent tasks, indicating strong capabilities in planning, problem-solving, and complex interaction flows.

Good For

  • Complex Chat Applications: Ideal for building chatbots that require nuanced responses and sophisticated conversational abilities.
  • Multilingual Communication: Suitable for applications needing to interact effectively in multiple languages.
  • Role-Playing and Creative Writing: Excels in scenarios demanding character emulation and imaginative text generation.
  • AI Agent Development: Provides strong foundational capabilities for developing AI agents that can perform intricate tasks and reasoning.

Popular Sampler Settings

Top 3 parameter combinations used by Featherless users for this model. Click a tab to see each config.

temperature
top_p
top_k
frequency_penalty
presence_penalty
repetition_penalty
min_p