AuriAetherwiing/G4-31B-Musica-v1
AuriAetherwiing/G4-31B-Musica-v1 is a 31 billion parameter Gemma-4-it based model fine-tuned for roleplay, story generation, and conversational tasks. It is the second model in the Musica series, offering improved instruction following and steerability over its Qwen-based predecessor. This model demonstrates strong performance in both reasoning and non-reasoning tasks, with non-existent refusals and enhanced swipe diversity.
Loading preview...
Model Overview
AuriAetherwiing/G4-31B-Musica-v1 is a 31 billion parameter model built upon the Gemma-4-31B-it architecture, specifically fine-tuned for roleplay, story generation, and general conversational applications. It represents the second iteration in the 'Musica' series, aiming to provide a significant upgrade over previous Qwen-based versions in its target domains.
Key Capabilities & Features
- Enhanced Roleplay & Story Generation: Optimized for creative text generation, offering improved narrative coherence and character interaction.
- Superior Instruction Following: Demonstrates high steerability and adherence to user instructions, with non-existent refusals.
- Reasoning Support: Capable of both reasoning and non-reasoning tasks, with support for DeepSeek-esque reasoning styles when pre-filled appropriately.
- Improved Diversity: Offers better 'swipe diversity' compared to the base Gemma model, leading to more varied and interesting outputs.
- Robust Training: Fine-tuned using a pretokenized dataset (
allura-forge/musica-sft-v1-gemma4-pretok) with a sequence length of 8192, employing a r64a64 LoRA configuration over one epoch.
When to Use This Model
This model is particularly well-suited for applications requiring:
- Engaging and creative roleplay scenarios.
- Long-form story generation and narrative development.
- Highly steerable conversational agents that follow complex instructions.
- Tasks where both reasoning and creative text generation are beneficial.