hungpill/Qwen3-0.6B-JSON-SFT-GRPO

TEXT GENERATIONPricing:Input $0.04 / Cached $0.008 / Output $0.08Concurrent Unit Cost:1Model Size:0.8BQuant:BF16Context Size:32kTool Calling:SupportedPublished:Sep 4, 2026License:apache-2.0Architecture:Transformer Open Weights Featherless Exclusive Cold

The hungpill/Qwen3-0.6B-JSON-SFT-GRPO is an 0.8 billion parameter Qwen3 model developed by hungpill, fine-tuned from NotoriousH2/Qwen3-0.6B-JSON-SFT. This model was trained using Unsloth and Huggingface's TRL library, achieving 2x faster training speeds. It is designed for tasks requiring a compact yet efficient language model, particularly benefiting from its optimized training process.

Loading preview...

Model Overview

The hungpill/Qwen3-0.6B-JSON-SFT-GRPO is an 0.8 billion parameter language model developed by hungpill. It is fine-tuned from the NotoriousH2/Qwen3-0.6B-JSON-SFT base model, leveraging the Qwen3 architecture. A key differentiator for this model is its optimized training process, which was conducted using Unsloth and Huggingface's TRL library, resulting in a reported 2x faster training speed.

Key Capabilities

  • Efficient Training: Achieves significantly faster training times due to the integration of Unsloth and TRL library optimizations.
  • Compact Size: With 0.8 billion parameters, it offers a balance between performance and computational efficiency.
  • Qwen3 Architecture: Benefits from the underlying capabilities of the Qwen3 model family.

Good For

  • Applications requiring a smaller, faster-to-train language model.
  • Scenarios where rapid iteration and deployment of fine-tuned models are crucial.
  • Developers looking for an efficient Qwen3-based model for various language tasks.