qwen3-4b-structeval-day8-clean

This adapter is fine-tuned from Qwen3-4B using Unsloth (QLoRA). Day 8 Special: Trained with strict data cleaning to remove markdown blocks.

Configuration

  • Epochs: 1
  • Learning Rate: 2e-4
  • Rank: 16
  • System Prompt: Enforced ('Output ONLY raw JSON code...')

Usage

See base model instructions.

Downloads last month
-
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Model tree for FumihiroToko/qwen3-4b-structeval-day8-clean

Adapter
(5261)
this model

Dataset used to train FumihiroToko/qwen3-4b-structeval-day8-clean