Upload moondream.py

moondream.py

@@ -84,10 +84,10 @@ class Moondream(PreTrainedModel):
             output_ids = self.text_model.generate(
                 inputs_embeds=inputs_embeds, streamer=streamer, **generate_config
             )
+            print("OUTPUTIDS" + output_ids)
 
         return tokenizer.batch_decode(output_ids, skip_special_tokens=True)
 
-    """
     def answer_question(
         self,
         image_embeds,
@@ -112,29 +112,6 @@ class Moondream(PreTrainedModel):
             result_queue.put(cleaned_answer)
         else:
             return cleaned_answer
-    """
-    async def answer_question(
-        self,
-        image_embeds,
-        question,
-        tokenizer,
-        chat_history="",
-        result_queue=None,
-        **kwargs,
-    ):
-        prompt = f"<image>\n\n{chat_history}Question: {question}\n\nAnswer:"
-        streamer = TextStreamer(tokenizer)
-
-
-        output_ids = self.text_model.generate(
-            inputs_embeds=self.input_embeds(prompt, image_embeds, tokenizer),
-            attention_mask=torch.ones(self.inputs_embeds.shape, dtype=torch.long, device=self.device), #maybe remove
-            streamer=streamer,
-            **kwargs,
-        )
-
-        for output_id in output_ids:
-            yield tokenizer.decode(output_id, skip_special_tokens=True)
 
     def batch_answer(
         self,
@@ -200,4 +177,4 @@ class Moondream(PreTrainedModel):
         return [
             x.strip()
             for x in tokenizer.batch_decode(output_ids, skip_special_tokens=True)
-        ]
+        ]