Update app.py

app.py

@@ -54,14 +54,14 @@ with open("bad_words.txt", "r") as f:
 model, image_processor, tokenizer = create_model_and_transforms(
     clip_vision_encoder_pretrained="openai",
     clip_vision_encoder_path="ViT-L-14",
-    lang_encoder_path="anas-awadalla/mpt-1b-redpajama-200b-dolly",
-    tokenizer_path="anas-awadalla/mpt-1b-redpajama-200b-dolly",
+    lang_encoder_path="anas-awadalla/mpt-7b",
+    tokenizer_path="anas-awadalla/mpt-7b",
     cross_attn_every_n_layers=1,
 )
 
-checkpoint_path = hf_hub_download("openflamingo/OpenFlamingo-3B-vitl-mpt1b-langinstruct", "checkpoint.pt")
+checkpoint_path = hf_hub_download("openflamingo/OpenFlamingo-9B-vitl-mpt7b", "checkpoint.pt")
 model.load_state_dict(torch.load(checkpoint_path), strict=False)
-
+model = model.to("cuda")
 model.eval()
 
 def generate(
@@ -153,9 +153,9 @@ def generate(
 
     # with torch.cuda.amp.autocast(dtype=torch.bfloat16):
     output = model.generate(
-        vision_x=vision_x,
-        lang_x=input_ids,
-        attention_mask=attention_mask,
+        vision_x=vision_x.to("cuda"),
+        lang_x=input_ids.to("cuda"),
+        attention_mask=attention_mask.to("cuda"),
         max_new_tokens=30,
         num_beams=3,
         # do_sample=True,