update chat

multimodal/open_flamingo/chat/conversation.py

@@ -361,8 +361,8 @@ class Chat:
             return_tensors="pt",
             max_length=2000,
         )
-        input_ids = encodings["input_ids"].to("cuda")
-        attention_mask = encodings["attention_mask"].to("cuda")
+        input_ids = encodings["input_ids"]
+        attention_mask = encodings["attention_mask"]
         image_start_index_list = ((input_ids == media_token_id).nonzero(as_tuple=True)[-1] + 1).tolist()
         image_start_index_list = [[x] for x in image_start_index_list]
         image_nums = [1] * len(input_ids)
@@ -466,7 +466,7 @@ def evaluate_exp(
     image = Image.open(image_path).convert("RGB")
     image = image.resize((size, size))
     print(f"image size: {image.size}")
-    batch_images = preprocess_image(image, image_processor).unsqueeze(0).unsqueeze(1).unsqueeze(0).to("cuda")
+    batch_images = preprocess_image(image, image_processor).unsqueeze(0).unsqueeze(1).unsqueeze(0)
     conversation = []
     human_sentence = None
     while True: