Sign up

TeichAI/Qwen3-4B-Instruct-2507-Polaris-Alpha-Distill

🤗 On Hugging Facetext-generationapache-2.04B params8.0 GBsafetensors✓ HF checksums availableupdated today
Magnet

Qwen3 4B Instruct 2507 x Polaris Alpha

This is a non-reasoning model trained on 1,000 examples from Polaris Alpha, an early snapshot of GPT-5.1 with reasoning effort set to minimal.

GGUFs available here

  • Developed by: TeichAI
  • License: apache-2.0
  • Finetuned from model : unsloth/Qwen3-4B-Instruct-2507

This qwen3 model was trained 2x faster with Unsloth and Huggingface's TRL library.

[](https://github.com/unslothai/unsloth)