styledrop / open_clip /model_configs /mt5-base-ViT-B-32.json
zideliu's picture
StyleDrop init
28c6826
{
"embed_dim": 512,
"vision_cfg": {
"image_size": 224,
"layers": 12,
"width": 768,
"patch_size": 32
},
"text_cfg": {
"hf_model_name": "google/mt5-base",
"hf_tokenizer_name": "google/mt5-base",
"proj": "mlp",
"pooler_type": "mean_pooler"
}
}