Monibee-Fudgekins/gemma-coder

VISIONConcurrent Unit Cost:2Model Size:26BQuant:FP8Context Size:32kTool Calling:SupportedPublished:Jun 29, 2026License:gemmaArchitecture:Transformer Featherless Exclusive Cold

Monibee-Fudgekins/gemma-coder is a 26 billion parameter coding-focused fine-tune of the Google Gemma 4 26B A4B-it model, an MoE with approximately 4 billion active parameters. Developed by Monibee-Fudgekins, this model is specifically optimized for multi-language coding tasks, ensuring performance meets target benchmarks and avoids regression against its base model. It was trained using Unsloth and Huggingface's TRL library, offering a specialized solution for code generation and understanding.

Loading preview...

Monibee-Fudgekins/gemma-coder Overview

Monibee-Fudgekins/gemma-coder is a specialized 26 billion parameter language model, fine-tuned from google/gemma-4-26B-A4B-it. This base model is a Mixture-of-Experts (MoE) architecture with approximately 4 billion active parameters, making it efficient for its size. Developed by Monibee-Fudgekins, this iteration is specifically engineered for coding-related tasks.

Key Capabilities & Features

  • Coding-focused Fine-tune: Optimized for generating and understanding code across multiple programming languages.
  • Performance Guarantee: Promotions to this repository are gated by an evaluation process, ensuring that the fine-tuned model's pass@1 metric meets specific targets and does not regress compared to the frozen base model on a sandboxed multi-language coding suite.
  • Efficient Training: The model was trained 2x faster utilizing Unsloth and Huggingface's TRL library, indicating an efficient and streamlined development process.
  • Context Length: Inherits a context length of 32768 tokens, allowing for processing of substantial code blocks.

Ideal Use Cases

This model is particularly well-suited for developers and applications requiring:

  • Code Generation: Creating new code snippets or functions.
  • Code Completion: Assisting developers by suggesting code as they type.
  • Code Understanding: Analyzing and interpreting existing codebases.
  • Multi-language Development: Projects involving various programming languages, given its multi-language coding suite evaluation.

Its rigorous evaluation gate ensures reliability for coding applications where performance and accuracy are critical.

Finetunes
1 models