trained models

README.md

@@ -1,5 +1,26 @@
 ---
 license: mit
-library_name: open_clip
-pipeline_tag: zero-shot-image-classification
----
+---
+
+# [Visual Product Recognition Challenge](https://www.aicrowd.com/challenges/visual-product-recognition-challenge-2023)
+
+
+The trained models for the competition. The training code for the models can be found in [HCA97/Product-Recognition](https://github.com/HCA97/Product-Recognition).
+
+
+# How to use it?
+
+You need to install `open_clip` library.
+
+```bash
+pip install open_clip
+```
+
+Example of loading the model:
+
+```py
+model = open_clip.create_model_and_transforms('ViT-H-14', None)[0].visual
+model.load_state_dict(th.load('path to model'))
+model.half()
+model.eval()
+```

model1.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:13d24332c79fb809ab40a9af19e64e018b428a918f5539324a4815c4685a5ff4
+size 1264309243

model2.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:729770f8e62c2742a5d1be6e4e1078cd62de89229f78d1563dc8f0268a8d585c
+size 1264312001

model3.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:11617b481eb495124b62a700ffcfe743739ef74045295712396e7cda134f76da
+size 1264297689