Aratako/gemma-3-4b-it-RP-v0.1
Aratako/gemma-3-4b-it-RP-v0.1 is a 4.3 billion parameter Gemma 3 instruction-tuned model developed by Aratako, fine-tuned specifically for role-playing scenarios. This model excels at generating character-driven dialogues and narratives based on detailed user-provided settings. With a context length of 32768 tokens, it is optimized for immersive and extended role-play interactions.
Loading preview...
Overview
Aratako/gemma-3-4b-it-RP-v0.1 is a 4.3 billion parameter model based on Google's gemma-3-4b-it architecture. It has been specifically fine-tuned by Aratako for role-playing (RP) applications, focusing on generating character-consistent responses within defined scenarios.
Key Capabilities
- Role-Play Optimization: The model is fine-tuned to excel in role-playing, allowing users to define character settings, worldviews, and dialogue tones.
- Detailed Scenario Handling: It can process extensive user prompts that include character descriptions, scene settings, and interaction styles to maintain narrative coherence.
- Context Length: Supports a substantial context length of 32768 tokens, enabling longer and more complex role-play sessions.
- Instruction Following: Designed to follow detailed instructions provided at the beginning of the user prompt, as Gemma 3 does not natively support system prompts.
Usage and Limitations
Users should provide all role-play settings and character details at the beginning of the initial user prompt. The model's training focused solely on text data, and its behavior with image inputs remains untested. The model inherits the Gemma Terms of Use and Prohibited Use Policy.
Training Details
The fine-tuning process utilized a learning rate of 1e-5 with a cosine scheduler, a global batch size of 128, and a maximum sequence length of 8192. An AdamW optimizer was used with a weight decay of 0.01.