temporary until tensor loading can accurately account for vision models
Attach a file by drag & drop or click to upload