nexus-flash-9B

Nexus-Flash-9B

This model is a fine-tuned version of unsloth/Qwen3.5-9B, optimized for agent-based reasoning tasks. It was trained using the Unsloth framework to achieve faster training speeds and memory efficiency.

📋 Model Details

• Developed by: ethanzxv
• Base Model: unsloth/Qwen3.5-9B
• License: Apache-2.0
• Language: English
• Finetuning Dataset: lambda/hermes-agent-reasoning-traces

🚀 Training & Optimization

This model was trained 2x faster using Unsloth combined with Hugging Face's TRL library. Unsloth allows for efficient fine-tuning of Large Language Models (LLMs) with significantly reduced VRAM usage and increased throughput.

<img src="https://raw.githubusercontent.com/unslothai/unsloth/main/images/unsloth%20made%20with%20love.png" width="200"/>

Dataset Information

The model was fine-tuned on the Hermes Agent Reasoning Traces dataset. This dataset focuses on enhancing the model's ability to perform complex reasoning steps, particularly in agentic workflows, by providing detailed traces of thought processes and decision-making paths.

• Dataset: lambda/hermes-agent-reasoning-traces

🎯 Intended Use & Capabilities

This model is designed for:

• Agent Reasoning: Improved performance in tasks requiring multi-step logical deduction.
• Complex Problem Solving: Better handling of intricate queries that require chain-of-thought processing.
• General Text Generation: Maintains the strong general capabilities of the base Qwen3.5-9B model.

📄 License

This model is released under the Apache-2.0 license. Please refer to the base model's license and the dataset's license for any additional restrictions or requirements.

🙏 Acknowledgements

• Unsloth: For providing the efficient fine-tuning framework.
• Hugging Face TRL: For the training reinforcement library.
• Lambda: For curating the Hermes Agent Reasoning Traces dataset.
• Alibaba Cloud: For the original Qwen3.5 base model.