lokeshe09/gemma-4-26B-A4B-it-GRPO-Math-16bit

VISIONConcurrency Cost:2Model Size:26BQuant:FP8Ctx Length:32kTool Calling:SupportedPublished:Jun 9, 2026License:apache-2.0Architecture:Transformer Open Weights Cold

The lokeshe09/gemma-4-26B-A4B-it-GRPO-Math-16bit is a 26 billion parameter Gemma 4 model, fine-tuned by lokeshe09. This instruction-tuned model was developed using Unsloth and Huggingface's TRL library, enabling faster training. It is designed for general conversational and instruction-following tasks, leveraging its large parameter count and efficient training methodology.

Loading preview...

Model Overview

The lokeshe09/gemma-4-26B-A4B-it-GRPO-Math-16bit is a 26 billion parameter instruction-tuned model based on the Gemma 4 architecture. Developed by lokeshe09, this model was fine-tuned from unsloth/gemma-4-26B-A4B-it.

Key Characteristics

  • Architecture: Gemma 4, a powerful open-source model family.
  • Parameter Count: 26 billion parameters, offering strong language understanding and generation capabilities.
  • Training Efficiency: Fine-tuned using Unsloth and Huggingface's TRL library, which facilitates 2x faster training.
  • Context Length: Supports a context window of 32768 tokens, allowing for processing longer inputs and generating more coherent extended responses.

Intended Use Cases

This model is suitable for a variety of instruction-following and conversational AI applications. Its large parameter size and instruction-tuned nature make it effective for:

  • General-purpose text generation.
  • Answering questions and providing information.
  • Engaging in dialogue and conversational agents.
  • Tasks requiring understanding and adherence to specific instructions.

Finetunes
3 models