update app.py

app.py

@@ -16,19 +16,28 @@ from PIL import Image
 from huggingface_hub import hf_hub_download, login
 
 from open_flamingo.src.factory import create_model_and_transforms
+
+
 flamingo, image_processor, tokenizer, vis_embed_size = create_model_and_transforms(
-        "ViT-L-14",
-        "datacomp_xl_s13b_b90k",
-        "EleutherAI/pythia-1.4b",
-        "EleutherAI/pythia-1.4b",
-        add_visual_grounding=True,
-        location_token_num=1000,
-        add_visual_token = True,
-        use_format_v2 = True,
-    )
+    args.vision_encoder_path,
+    args.vision_encoder_pretrained,
+    args.lm_path,
+    args.lm_tokenizer_path,
+    location_token_num=1000,
+    lora=False,
+    lora_r=16,
+    use_sam=None,
+    add_visual_token=True,
+    use_format_v2=True,
+    add_box=True,
+    add_pe=False,
+    add_relation=False,
+    enhance_data=False,
+)
+
 
 checkpoint_path = hf_hub_download("chendl/compositional_test", "pythiaS.pt")
-checkpoint = torch.load(checkpoint_path, map_location="cpu")
+checkpoint = torch.load(checkpoint_path, map_location="cpu")["model_state_dict"]
 model_state_dict = {}
 for key in checkpoint.keys():
     model_state_dict[key.replace("module.", "")] = checkpoint[key]