YAML Metadata Warning:empty or missing yaml metadata in repo card

Check out the documentation for more information.

0927_v1

Qwen3.5-9B full-parameter SFT checkpoint-200 (not a LoRA adapter), converted to BF16 for inference. Original FP32 checkpoint preserved separately. Training: 430 new-data images + 144 historical July examples. This is NOT the full-history 0927_v2.

Validation before export: macro image-class recall 0.6345, micro image-class recall 0.7462 on 92 positive images. These are class-presence metrics, not localized instance recall; false positives on normal images are not evaluated. See inference_prompt.txt. Training data and optimizer state are intentionally excluded.

Downloads last month
12
Safetensors
Model size
9B params
Tensor type
BF16
·
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support