lunahr/thea-rp-3b-25r
TEXT GENERATIONConcurrency Cost:1Model Size:3.2BQuant:BF16Ctx Length:32kPublished:Oct 13, 2024License:llama3.2Architecture:Transformer0.0K Cold
lunahr/thea-rp-3b-25r is an uncensored 3.2 billion parameter Llama 3.2 model developed by Piotr Zalewski, specifically fine-tuned for roleplay and reasoning tasks. It was trained on reasoning data using custom, optimized training code, potentially achieving high scores for roleplay fine-tunes of Llama 3.2. This model is designed for applications requiring robust reasoning capabilities within a roleplay context.
Loading preview...
Overview
lunahr/thea-rp-3b-25r is a 3.2 billion parameter Llama 3.2 model, developed by Piotr Zalewski, that has been specifically fine-tuned for uncensored roleplay and reasoning. It builds upon the SicariusSicariiStuff/Impish_LLAMA_3B base model and was trained using the KingNish/reasoning-base-20k dataset.
Key Capabilities
- Uncensored Roleplay: Designed to generate responses suitable for diverse roleplay scenarios without content restrictions.
- Enhanced Reasoning: Incorporates training on reasoning data, which is intended to improve its logical processing and problem-solving abilities.
- Optimized Training: Utilizes custom training code, which is noted to be faster than Unsloth, potentially leading to improved performance for its size.
Good for
- Applications requiring a compact (3.2B parameters) yet capable model for roleplay generation.
- Use cases where reasoning abilities are crucial within a conversational or narrative context.
- Developers seeking an uncensored model for flexible content generation in roleplaying applications.