Target: 800 approved cases to fine-tune Mistral 7B
0 / 800
0200400600800 ✓ Ready
100
Cases
Initial experiments
300
Cases
First test fine-tune
600
Cases
Solid domain coverage
800
Cases
🚀 Fine-tune ready!
⬇ Export Training Dataset
Export your approved cases in the correct format for model fine-tuning. JSONL is required for Mistral 7B and LLaMA 3.
JSONL Format — Required for fine-tuning Mistral 7B, LLaMA 3 via Hugging Face or Axolotl. Each line = one training pair. Only approved (👍) cases are exported.