Zheng-Zong/AronaR1-SFT-stage1-v2

TEXT GENERATIONConcurrency Cost:1Model Size:7.6BQuant:FP8Ctx Length:32kPublished:Mar 15, 2026License:apache-2.0Architecture:Transformer Open Weights Cold

Zheng-Zong/AronaR1-SFT-stage1-v2 is a Qwen2-based instruction-tuned language model developed by Zheng-Zong. This model was finetuned from unsloth/Qwen2.5-Math-7B-Instruct, leveraging Unsloth and Huggingface's TRL library for accelerated training. It is optimized for specific tasks related to its mathematical instruction-tuned base model, offering enhanced performance in those domains. The model is released under the Apache-2.0 license.

Loading preview...

Model Overview

Zheng-Zong/AronaR1-SFT-stage1-v2 is an instruction-tuned language model developed by Zheng-Zong. It is built upon the Qwen2 architecture and was specifically finetuned from the unsloth/Qwen2.5-Math-7B-Instruct base model. This finetuning process utilized Unsloth and Huggingface's TRL library, which enabled a 2x faster training speed.

Key Characteristics

  • Base Model: Finetuned from unsloth/Qwen2.5-Math-7B-Instruct, indicating a foundation optimized for mathematical reasoning.
  • Training Efficiency: Leverages Unsloth for accelerated training, suggesting a focus on efficient model development.
  • License: Distributed under the Apache-2.0 license, allowing for broad use and modification.

Potential Use Cases

Given its finetuning from a math-oriented base model, AronaR1-SFT-stage1-v2 is likely suitable for applications requiring:

  • Mathematical problem-solving: Assisting with calculations, equations, and logical reasoning in mathematical contexts.
  • Instruction following: Performing tasks based on specific instructions, particularly those related to its training domain.
  • Efficient deployment: Models trained with Unsloth often benefit from optimized inference, making them suitable for resource-constrained environments.