D1rtyB1rd/Looking-Glass-Alice-Thinking-NSFW-RP-8B
D1rtyB1rd/Looking-Glass-Alice-Thinking-NSFW-RP-8B is a fine-tuned 8-billion parameter Llama 3-based language model, derived from DoppelReflEx/L3-8B-R1-WolfCore. This model was specifically trained for 'thinking' processes, though it may exhibit some repetitiveness. It is intended for use with high temperature and repetition penalty settings, and is suitable for roleplay scenarios.
Loading preview...
Overview
D1rtyB1rd/Looking-Glass-Alice-Thinking-NSFW-RP-8B is an 8-billion parameter language model based on the Llama 3 architecture, fine-tuned from DoppelReflEx/L3-8B-R1-WolfCore. This model was developed as an interim version for testing and further planned training, with a focus on enhancing its 'thinking' capabilities. It was trained using the TRL (Transformer Reinforcement Learning) framework.
Key Characteristics
- Base Model: Fine-tuned from DoppelReflEx/L3-8B-R1-WolfCore, which is a Llama 3-based model.
- Training Focus: Specifically trained for 'thinking' processes.
- Known Behavior: The model may exhibit some repetitiveness in its outputs.
- Recommended Usage: Users are advised to employ a high temperature and repetition penalty for optimal results.
- Template: Uses the Llama 3 template for prompt formatting.
Use Cases
- Roleplay (RP): The model is suitable for roleplay applications, particularly those involving 'thinking' or internal monologue aspects.
- Experimental Development: Given its interim status, it can be used for testing and exploring fine-tuned Llama 3 behaviors, especially concerning thought generation.