allura-org/gemma-4-12B-it-blorbo-v0b

TEXT GENERATIONConcurrent Unit Cost:1Model Size:12BQuant:FP8Context Size:32kTool Calling:SupportedPublished:Jul 18, 2026License:apache-2.0Architecture:Transformer0.0K Open Weights Featherless Exclusive Cold

allura-org/gemma-4-12B-it-blorbo-v0b is a 12 billion parameter instruction-tuned Gemma 4 model developed by allura-org, fine-tuned for improved reasoning and human-like roleplaying/story generation. It leverages a 32,768 token context length and incorporates specific training on reasoning data from Mimo v2.5 Pro and Mimo v2 Pro, alongside human roleplaying datasets. This model is designed to consistently handle both reasoning tasks and creative narrative generation, making it suitable for applications requiring nuanced conversational abilities.

Loading preview...

Model Overview

allura-org/gemma-4-12B-it-blorbo-v0b is a 12 billion parameter instruction-tuned variant of the Gemma 4 model, developed by allura-org. This model has been specifically fine-tuned to enhance its capabilities in both reasoning and human-like roleplaying/story generation.

Key Capabilities & Training

  • Reasoning: The model was trained on reasoning data generated with Mimo v2.5 Pro and Mimo v2 Pro. While Doubao Seed 2.0 Pro reasoning data was included, its reasoning style was masked, leading the model to adopt Mimo's typically shorter reasoning style.
  • Roleplaying & Story Generation: It incorporates extensive training on human roleplaying and story datasets, allowing it to consistently generate creative narratives.
  • Architecture: Based on Gemma 4, it maintains a 32,768 token context length. The fine-tuning process involved untied embeddings and lm_head layers, optimized with ScheduleFree AdamW and R64/A512 16-bit LoRA.

Usage Considerations

  • Format: The model uses the standard Gemma 4 format, with <|think|> still functioning to trigger reasoning processes.
  • Sampler Settings: Recommended sampler settings include a temperature range of 1.0-1.25 with either 0.1 min_p or 0.95 top_p.

Potential Use Cases

  • Interactive Storytelling: Generating dynamic and engaging narratives based on user prompts.
  • Roleplaying Scenarios: Creating consistent and believable character interactions.
  • Reasoning-based Chatbots: Applications requiring logical deduction within conversational contexts, particularly where concise reasoning is preferred.

Finetunes
1 models