|
|
--- |
|
|
license: llama3 |
|
|
language: |
|
|
- en |
|
|
tags: |
|
|
- roleplay |
|
|
- storytelling |
|
|
- language-model |
|
|
--- |
|
|
|
|
|
# T-Rex-mini |
|
|
|
|
|
T-Rex-mini is an 8B parameter language model specifically optimized for roleplaying and storytellin based on Llama-3 Instruct. This model is the result of merging several high-quality roleplay models to create an efficient and powerful storytelling assistant that can run locally on consumer hardware. |
|
|
|
|
|
## Key Features |
|
|
|
|
|
- Optimized for creative writing and character-based interactions |
|
|
- Compact 8B parameter size for local usage |
|
|
- Built on the same training foundation as our production models at [LoreMate](https://loremate.saturated.in) |
|
|
- Fine-tuned for engaging narrative generation and character dialogue. |
|
|
|
|
|
## Please Support Us |
|
|
|
|
|
We are running our platform for free, and training these models costs us crazy money so we'd really appricaite your support via ko-fi: [Donation Link](https://ko-fi.com/loremate) |
|
|
|
|
|
|
|
|
|
|
|
## Model Details |
|
|
|
|
|
This open-source model aims to make advanced roleplaying AI accessible to the community. It provides a balance between performance and resource requirements, making it ideal for personal use and experimentation. |
|
|
|
|
|
Settings: |
|
|
``` |
|
|
Temperature: 1.35 |
|
|
top_p: 1.0 |
|
|
min_p: 0.1 |
|
|
presence_penalty: 0.0 |
|
|
frequency_penalty: 0.0 |
|
|
repetition_penalty: 1.0 |
|
|
``` |
|
|
|
|
|
--- |
|
|
|
