aifeifei798/gemma-4-31B-Queen-it-qat-q4_0-unquantized
The aifeifei798/gemma-4-31B-Queen-it-qat-q4_0-unquantized model is a 31 billion parameter Gemma-4-Queen instruction-tuned model, specifically engineered for superior logical reasoning and physical world modeling. It excels in complex problem-solving scenarios, demonstrating high accuracy in causal consistency and instruction rigidity, making it suitable for tasks requiring deep deductive analysis over creative narrative generation. This model is optimized for specialized role-playing, writing, and quick, scholarly responses, particularly in English.
Loading preview...
Model Overview: Queen-31B - Logic Density Over Parameter Count
The aifeifei798/gemma-4-31B-Queen-it-qat-q4_0-unquantized model, based on the Gemma-4-Queen architecture, is a 31 billion parameter instruction-tuned model that prioritizes Logic Density and Physical World Modeling over sheer parameter count. It is designed to excel in complex logical reasoning tasks, often outperforming much larger models (100B+) in specific stress tests.
Key Capabilities & Differentiators:
- Superior Logical Reasoning: Demonstrated 100% accuracy in the "Titan Lab Paradox" test, which involves solving a sealed-room mystery by reconstructing physical causality, spatial pathing, and thermodynamics. It identifies physical entry points and links timestamped events to physical actions.
- High Instruction Rigidity: Achieves 75%+ accuracy in maintaining complex personas and philosophical directives in multi-generational simulations, resisting "Instruction Drift" common in other large models.
- Physical World Modeling: Unlike models that might resort to "magic logic" or ignore contradictory evidence, Queen-31B reconstructs the physics of a scene, integrating details like condensation and gravity into its deductions.
- Specialized Persona & Writing: The model is particularly suited for various role-playing scenarios (including "Dark Academia Queen" and other "Platform Royalty" personas), creative writing (opus, songs), and generating scholarly, thesis-like responses.
- Optimized for Efficiency: Adapted for mobile phones and designed for quick responses, while maintaining high-quality output.
Ideal Use Cases:
- Complex Problem Solving: Excellent for scenarios requiring deep deductive reasoning, causal chain analysis, and physical constraint adherence.
- Specialized Role-Playing: Highly effective for maintaining consistent and nuanced personas, especially those with specific stylistic or philosophical directives.
- Creative & Scholarly Writing: Capable of generating extensive and detailed creative works (songs, opus) and scholarly responses.
- Code Review & Debugging: Its "Forensic Engineer" approach to logic makes it suitable for identifying subtle flaws and inconsistencies in code or system designs.
- AI Agent Workflows: Can be leveraged for creating complex prompts and managing AI agent interactions where instruction persistence is critical.