AliceThirty/G4-Alice-v1.0-31B
G4-Alice-v1.0-31B is a 31 billion parameter language model developed by AliceThirty, based on a fine-tuned Serenity model. It is specifically designed to be less censored and more "uninhibited" compared to other models like MeroMero, aiming for a distinct personality. This model focuses on generating creative and less constrained text, making it suitable for applications requiring highly expressive and unfiltered responses.
Loading preview...
G4-Alice-v1.0-31B: An Uninhibited 31B Parameter Model
G4-Alice-v1.0-31B is a 31 billion parameter language model developed by AliceThirty, representing their first fine-tuning attempt. This model was created with the explicit goal of producing less censored and more "horny" or uninhibited outputs, contrasting with models like MeroMero. It is based on the Serenity-31B model, which was used as the base for retraining after initial merging attempts proved unsuccessful.
Key Characteristics
- Uninhibited Output: Designed to be less censored and more expressive, aiming for a distinct, unconstrained personality.
- Base Model: Fine-tuned from the Serenity-31B model.
- Training Method: Fine-tuned on a single RTX 5090 using unsloth, with a LoRA rank of 80 for one epoch and a low learning rate to preserve Serenity's characteristics.
- Parameter Count: 31 billion parameters.
- Context Length: Supports a context length of 32768 tokens.
Potential Use Cases
- Creative Writing: Ideal for generating highly imaginative, unfiltered, or adult-themed narratives and dialogues.
- Roleplay Scenarios: Suitable for roleplaying applications where less constrained and more expressive character interactions are desired.
- Exploratory Content Generation: Useful for developers experimenting with models that push boundaries in terms of content and tone.