Upload app.py

app.py

@@ -23,7 +23,7 @@ device = "cuda" if torch.cuda.is_available() else "cpu"
 feature_extractor = CLIPFeatureExtractor.from_pretrained("openai/clip-vit-base-patch32")
 
 # load and configure tokenizer
-tokenizer = AutoTokenizer.from_pretrained("facebook/opt-125m")
+tokenizer = AutoTokenizer.from_pretrained("facebook/opt-125M")
 tokenizer.pad_token = '!'
 tokenizer.eos_token = '.'
 
@@ -48,9 +48,9 @@ model= model.to(device)
 template = open('src/template.txt').read().strip() + ' '
 
 # precompute captions for retrieval
-captions = json.load(open('coco_index_captions.json'))
+captions = json.load(open('datastore/coco_index_captions.json'))
 retrieval_model, feature_extractor_retrieval = clip.load("RN50x64", device=device)
-retrieval_index = faiss.read_index('coco_index')
+retrieval_index = faiss.read_index('datastore/coco_index')
 #res = faiss.StandardGpuResources()  
 #retrieval_index = faiss.index_cpu_to_gpu(res, 0, retrieval_index)
 
@@ -99,4 +99,4 @@ textbox = gr.Textbox(placeholder="Retrieved captions and generated caption...",
 title = "SmallCap Demo"
 gr.Interface(
     fn=classify_image, inputs=image, outputs=textbox, title=title
-).launch()
+).launch(share=True)