Models
Docs
Pricing
Sign in
Download
Models
Download
Docs
Pricing
Sign in
bsspl
/
zuraai-vl-v3
:latest
2
Downloads
Updated
19 hours ago
Fine-tuned vision-language model that reads photographed IDs, invoices, receipts, and certificates and returns clean structured JSON — 33 document types, 83% doc-type accuracy on held-out eval. Q4_K_M quantized for fast local inference.
Fine-tuned vision-language model that reads photographed IDs, invoices, receipts, and certificates and returns clean structured JSON — 33 document types, 83% doc-type accuracy on held-out eval. Q4_K_M quantized for fast local inference.
Cancel
vision
zuraai-vl-v3:latest
...
/
params
11612710a889 · 20B
{
"num_predict": 600
}