xl-24/gemma4-26b-atc-finetune_sp-merged-16bit

VISIONConcurrency Cost:2Model Size:26BQuant:FP8Ctx Length:32kTool Calling:SupportedPublished:May 31, 2026License:apache-2.0Architecture:Transformer0.0K Open Weights Cold

The xl-24/gemma4-26b-atc-finetune_sp-merged-16bit is a 26 billion parameter Gemma 4 model, fine-tuned by xl-24. This model was optimized for faster training using Unsloth and Huggingface's TRL library, offering a 32768 token context length. It is designed for efficient deployment and performance in generative AI tasks.

Loading preview...

Model Overview

The xl-24/gemma4-26b-atc-finetune_sp-merged-16bit is a 26 billion parameter language model, fine-tuned by xl-24. It is based on the Gemma 4 architecture and was specifically optimized for training efficiency.

Key Characteristics

  • Base Model: Fine-tuned from unsloth/gemma-4-26b-a4b-it.
  • Training Optimization: Leverages Unsloth and Huggingface's TRL library, resulting in a 2x faster training process.
  • Parameter Count: Features 26 billion parameters, balancing performance with computational requirements.
  • Context Length: Supports a substantial context window of 32768 tokens.

Use Cases

This model is suitable for applications requiring a powerful Gemma 4-based LLM that benefits from optimized training and efficient deployment. Its fine-tuned nature suggests potential for specialized tasks, while the 26B parameters and large context window make it versatile for various generative AI applications.