saturated-labs/T-Rex-mini

Hugging Face
TEXT GENERATIONConcurrency Cost:1Model Size:8BQuant:FP8Ctx Length:8kPublished:Apr 5, 2025License:llama3Architecture:Transformer0.0K Warm

T-Rex-mini is an 8 billion parameter language model from saturated-labs, based on Llama-3 Instruct. It is specifically optimized for roleplaying and storytelling, resulting from the merge of several high-quality roleplay models. This model is designed for efficient narrative generation and character dialogue, making it suitable for local usage on consumer hardware.

Loading preview...

T-Rex-mini: An 8B Model for Roleplaying and Storytelling

T-Rex-mini is an 8 billion parameter language model developed by saturated-labs, built upon the Llama-3 Instruct architecture. This model is specifically engineered through the merging of multiple high-quality roleplay models to excel in creative writing and character-based interactions. Its compact size allows for efficient local deployment on consumer hardware, making advanced roleplaying AI accessible to a wider community.

Key Capabilities

  • Optimized for Creative Writing: Fine-tuned for engaging narrative generation and dynamic character dialogue.
  • Efficient Local Usage: Designed with an 8B parameter count for performance on consumer-grade hardware.
  • LoreMate Foundation: Built on the same training principles as saturated-labs' production models at LoreMate.

Ideal Use Cases

  • Roleplaying Scenarios: Generating rich, character-driven interactions.
  • Storytelling Assistance: Aiding in the creation of compelling narratives.
  • Personal AI Experimentation: Suitable for individual users exploring advanced AI capabilities locally.

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