NICKO/vit-base-patch16-224-v12-fpq

ViT Base Patch16-224 — BonfyreFPQ v12

Native .fpq v12 quantized weights for google/vit-base-patch16-224.

Model Info

Property	Value
Architecture	Vision Transformer (Image Classification)
Parameters	86M
Original Size	330 MB (safetensors)
FPQ v12 Size	78 MB
Compression	4.2×
Avg Cosine	0.999733
Worst Cosine	0.998490

Format

BonfyreFPQ v12 native format: rANS entropy-coded E8 lattice coordinates with 6-bit packed tiles and FP16 scales.

Decode to safetensors

bonfyre-fpqx decode vit-base-patch16-224-v12-fpq.fpq output.safetensors

Source

Converted from google/vit-base-patch16-224 using BonfyreFPQ v12.