Update README.md

README.md

@@ -38,7 +38,7 @@ device = torch.device("cuda:0" if torch.cuda.is_available() else "cpu")
 
 # load checkpoint from local path
 # model_path = "path/to/open_clip_pytorch_model.bin"
-# model_name = "ViT-SO400M-14-SigLIP-384"
+# model_name = "ViT-B-16"
 # model, preprocess = create_model_from_pretrained(model_name=model_name, pretrained=model_path, device=device)
 # tokenizer = get_tokenizer(model_name)
 
@@ -64,18 +64,16 @@ with torch.autocast(device_type=device.type):
     with torch.no_grad():
         image_features = model.encode_image(image)
         text_features = model.encode_text(text)
-        image_features = F.normalize(image_features, dim=-1)
-        image_features = F.normalize(image_features, dim=-1)
-        text_features = F.normalize(text_features, dim=-1)
-        text_probs = torch.sigmoid(image_features @ text_features.T * model.logit_scale.exp() + model.logit_bias)
-        scores = [f"{s:.3f}" for i, s in enumerate(text_probs.tolist()[0])]
-        print(scores)  # [0.016, 0.000, 0.001, 0.233]
+        image_features /= image_features.norm(dim=-1, keepdim=True)
+        text_features /= text_features.norm(dim=-1, keepdim=True)
+        text_probs = (100.0 * image_features @ text_features.T).softmax(dim=-1)
+        print(text_probs) # [0.0319, 0.0062, 0.0012, 0.9608]
 ```
 
 ## Model Card
 ### CLIP for GenshinImpact
 
-[CLIP model](https://huggingface.co/laion/CLIP-ViT-B-16-laion2B-s34B-b88K) further fine-tuned on 17k Genshin Impact English text-image pairs at resolution 384x384.
+[CLIP-ViT-B-16-laion2B-s34B-b88K model](https://huggingface.co/laion/CLIP-ViT-B-16-laion2B-s34B-b88K) further fine-tuned on 17k Genshin Impact English text-image pairs at resolution 384x384.
 
 ### Training data description