Update app.py

app.py

@@ -47,32 +47,56 @@ def respond1(
             yield response
 def respond0(multimodal_input,password):
     if password==password1:
+        if multimodal_input["files"] ==not None:
         
-        images = multimodal_input["files"]
-        content = [{"type": "image"} for _ in images]
-        content.append({"type": "text", "text": multimodal_input["text"]})
-        messages=[{"role":"system", "content":[{"type":"text", "text":"Your name is Chatchat.And, your made by SungYoon.In Korean, 정성윤.And these are the instructions:"+"You are a helpful assietant."}]}]
-        messages.append([{"role": "user", "content": content}])
-        response = ""
-
-        model_id = "HuggingFaceM4/idefics2-8b"
-
-        quantization_config = BitsAndBytesConfig(load_in_4bit=True, bnb_4bit_quant_type="nf4", bnb_4bit_use_double_quant=True, bnb_4bit_compute_dtype=torch.float16)
-        processor = AutoProcessor.from_pretrained(model_id)
-        model = AutoModelForVision2Seq.from_pretrained(model_id, torch_dtype=torch.float16, quantization_config=quantization_config)
-        prompt = processor.apply_chat_template(messages, add_generation_prompt=True)
-        inputs = processor(text=prompt, images=[images], return_tensors="pt")
-        inputs = {k: v.to(model.device) for k, v in inputs.items()}
-        num_tokens = len(inputs["input_ids"][0])
-        with torch.inference_mode():
-            generated_ids = model.generate(**inputs, max_new_tokens=max_tokens,top_p=top_p, temperature=1.0,)
-
-        new_tokens = generated_ids[:, num_tokens:]
-        generated_text = processor.batch_decode(new_tokens, skip_special_tokens=True)[0]
-
-        token = generated_text
-        response+=token
-        yield response
+            images = multimodal_input["files"]
+            content = [{"type": "image"} for _ in images]
+            content.append({"type": "text", "text": multimodal_input["text"]})
+            messages=[{"role":"system", "content":[{"type":"text", "text":"Your name is Chatchat.And, your made by SungYoon.In Korean, 정성윤.And these are the instructions:"+"You are a helpful assietant."}]}]
+            messages.append([{"role": "user", "content": content}])
+            response = ""
+
+            model_id = "HuggingFaceM4/idefics2-8b"
+
+            processor = AutoProcessor.from_pretrained(model_id)
+            model = AutoModelForVision2Seq.from_pretrained(model_id, torch_dtype=torch.float16)
+            prompt = processor.apply_chat_template(messages, add_generation_prompt=True)
+            inputs = processor(text=prompt, images=[images], return_tensors="pt")
+            inputs = {k: v.to(model.device) for k, v in inputs.items()}
+            num_tokens = len(inputs["input_ids"][0])
+            with torch.inference_mode():
+                generated_ids = model.generate(**inputs, max_new_tokens=max_tokens,top_p=top_p, temperature=1.0,)
+
+            new_tokens = generated_ids[:, num_tokens:]
+            generated_text = processor.batch_decode(new_tokens, skip_special_tokens=True)[0]
+
+            token = generated_text
+            response+=token
+            yield response
+        else:
+            
+            content={"type": "text", "text": multimodal_input["text"]}
+            messages=[{"role":"system", "content":[{"type":"text", "text":"Your name is Chatchat.And, your made by SungYoon.In Korean, 정성윤.And these are the instructions:"+"You are a helpful assietant."}]}]
+            messages.append([{"role": "user", "content": content}])
+            response = ""
+
+            model_id = "HuggingFaceM4/idefics2-8b"
+
+            processor = AutoProcessor.from_pretrained(model_id)
+            model = AutoModelForVision2Seq.from_pretrained(model_id, torch_dtype=torch.float16)
+            prompt = processor.apply_chat_template(messages, add_generation_prompt=True)
+            inputs = processor(text=prompt, images=[images], return_tensors="pt")
+            inputs = {k: v.to(model.device) for k, v in inputs.items()}
+            num_tokens = len(inputs["input_ids"][0])
+            with torch.inference_mode():
+                generated_ids = model.generate(**inputs, max_new_tokens=max_tokens,top_p=top_p, temperature=1.0,)
+
+            new_tokens = generated_ids[:, num_tokens:]
+            generated_text = processor.batch_decode(new_tokens, skip_special_tokens=True)[0]
+
+            token = generated_text
+            response+=token
+            yield response