DavidAU/gemma-3-12b-it-vl-GLM-4.7-Flash-Heretic-Uncensored-Thinking

Overview

DavidAU/gemma-3-12b-it-vl-GLM-4.7-Flash-Heretic-Uncensored-Thinking is a 12 billion parameter instruction-tuned model based on the Gemma architecture, fine-tuned by DavidAU. It leverages the GLM 4.7 Flash reasoning dataset and Unsloth for training, focusing on delivering fully uncensored and deep thinking capabilities.

Key Capabilities

• Uncensored Output: Designed to provide direct responses without refusals, even for sensitive or explicit content, requiring minimal prompting to achieve desired graphic or explicit levels.
• Enhanced Reasoning: Integrates "deep thinking" capabilities that affect general model operation, output generation, image processing, and benchmarks. Users can optionally activate this with "think deeply: prompt".
• Extended Context Window: Features a substantial 128k context length, allowing for processing and generating longer, more complex interactions.
• Temperature Stability: Reasoning remains stable across a wide temperature range of 0.1 to 2.5.

Good For

• Unfiltered Content Generation: Ideal for applications requiring explicit, graphic, or sensitive content without built-in censorship.
• Detailed and Direct Responses: Suited for use cases where concise yet highly detailed and direct answers are paramount.
• Creative and Roleplay Scenarios: Particularly effective in chat and roleplay environments where uncensored and imaginative outputs are desired, with recommended settings like a Smoothing_factor of 1.5 for smoother operation.
• Image Processing Enhancement: Benefits from improved reasoning for tasks involving image processing.