Update app.py

app.py

@@ -136,32 +136,11 @@ def answer_question(img, prompt):
         yield "", ""
         return
 
-    image_embeds = moondream.encode_image(img)
-    streamer = TextIteratorStreamer(tokenizer, skip_special_tokens=True)
-    queue = Queue()
-    thread = Thread(
-        target=moondream.answer_question,
-        kwargs={
-            "image_embeds": image_embeds,
-            "question": prompt,
-            "tokenizer": tokenizer,
-            "allow_cot": True,
-            "result_queue": queue,
-            "streamer": streamer,
-        },
-    )
-    thread.start()
-
     buffer = ""
-    for new_text in streamer:
+    for new_text in moondream.query(img, prompt, stream=True)["answer"]:
         buffer += new_text
         yield buffer.strip(), {"text": "Thinking...", "entities": []}
 
-    answer = queue.get()
-    thought = convert_to_entities(answer["thought"], answer["coords"])
-
-    yield answer["answer"], thought
-
 
 @spaces.GPU(duration=10)
 def caption(img, mode):
@@ -170,7 +149,7 @@ def caption(img, mode):
         return
 
     buffer = ""
-    for t in moondream.caption(length="short" if mode == "Short" else None, stream=True)["caption"]:
+    for t in moondream.caption(img, length="short" if mode == "Short" else None, stream=True)["caption"]:
         buffer += t
         yield buffer.strip()