Update app.py

app.py

@@ -68,9 +68,9 @@ def encode_docs(docs,maxlen = 64, stride = 32):
     
     embeddings = np.float32(torch.stack(embeddings).transpose(0, 1).cpu())
     
-    np.save("encoded_gradio/emb_{}.npy".format(name),dict(zip(list(range(len(embeddings))),embeddings))) 
-    np.save("encoded_gradio/spans_{}.npy".format(name),dict(zip(list(range(len(spans))),spans)))
-    np.save("encoded_gradio/file_{}.npy".format(name),dict(zip(list(range(len(file_names))),file_names)))
+    np.save("emb_{}.npy".format(name),dict(zip(list(range(len(embeddings))),embeddings))) 
+    np.save("spans_{}.npy".format(name),dict(zip(list(range(len(spans))),spans)))
+    np.save("file_{}.npy".format(name),dict(zip(list(range(len(file_names))),file_names)))
     
     return embeddings, spans, file_names
    
@@ -81,16 +81,16 @@ def predict(query,data):
     now = datetime.now()
     current_time = now.strftime("%H:%M:%S")
     try:
-        df = pd.read_csv("HISTORY/{}.csv".format(hash(st)))
+        df = pd.read_csv("{}.csv".format(hash(st)))
         return df
     except Exception as e:
         print(e)
         print(st)
 
     if name_to_save+".txt" in os.listdir("text_gradio"):
-        doc_emb = np.load('encoded_gradio/emb_{}.npy'.format(name_to_save),allow_pickle='TRUE').item()
-        doc_text = np.load('encoded_gradio/spans_{}.npy'.format(name_to_save),allow_pickle='TRUE').item()
-        file_names_dicto = np.load('encoded_gradio/file_{}.npy'.format(name_to_save),allow_pickle='TRUE').item()
+        doc_emb = np.load('emb_{}.npy'.format(name_to_save),allow_pickle='TRUE').item()
+        doc_text = np.load('spans_{}.npy'.format(name_to_save),allow_pickle='TRUE').item()
+        file_names_dicto = np.load('file_{}.npy'.format(name_to_save),allow_pickle='TRUE').item()
         
         doc_emb = np.array(list(doc_emb.values())).reshape(-1,768)
         doc_text = list(doc_text.values())
@@ -105,7 +105,7 @@ def predict(query,data):
         doc_emb, doc_text, file_names = encode_docs((name_to_save,text),maxlen = 64, stride = 32)
         
         doc_emb = doc_emb.reshape(-1, 768)
-        with open("text_gradio/{}.txt".format(name_to_save),"w",encoding="utf-8") as f:
+        with open("{}.txt".format(name_to_save),"w",encoding="utf-8") as f:
             f.write(text)
     start = time.time()
     query_emb = encode_query(query)
@@ -154,7 +154,7 @@ def predict(query,data):
         f.write(" " + str(current_time))
         f.write("\n")
         f.close()
-    df.to_csv("HISTORY/{}.csv".format(hash(st)), index=False)
+    df.to_csv("{}.csv".format(hash(st)), index=False)
     
     return df
 
@@ -162,7 +162,6 @@ iface = gr.Interface(
     
     fn =predict,
     inputs = [gr.inputs.Textbox(default="What is Open-domain question answering?"),
-            gr.inputs.Checkbox(default=True),
               gr.inputs.File(),
     ],
     outputs = [