Update LLaVA-NeXT.md

docs/LLaVA-NeXT.md

@@ -57,30 +57,37 @@ from llava.mm_utils import get_model_name_from_path, process_images, tokenizer_i
 from llava.constants import IMAGE_TOKEN_INDEX, DEFAULT_IMAGE_TOKEN, DEFAULT_IM_START_TOKEN, DEFAULT_IM_END_TOKEN, IGNORE_INDEX
 from llava.conversation import conv_templates, SeparatorStyle
 
-pretrained = "https://huggingface.co/lmms-lab/llama3-llava-next-8b"
+from PIL import Image
+import requests
+import copy
+import torch
+
+pretrained = "lmms-lab/llama3-llava-next-8b"
 model_name = "llava_llama3"
-tokenizer, model, image_processor, max_length = load_pretrained_model(pretrained, None, model_name, device_map=self.device_map, **llava_model_args) # Add any other thing you want to pass in llava_model_args
+device = "cuda"
+device_map = "auto"
+tokenizer, model, image_processor, max_length = load_pretrained_model(pretrained, None, model_name, device_map=device_map) # Add any other thing you want to pass in llava_model_args
 
 model.eval()
 model.tie_weights()
 
 url = "https://github.com/haotian-liu/LLaVA/blob/1a91fc274d7c35a9b50b3cb29c4247ae5837ce39/images/llava_v1_5_radar.jpg?raw=true"
 image = Image.open(requests.get(url, stream=True).raw)
-image_tensor = process_images(flattened_visuals, image_processor, model.config).to(dtype=torch.float16, device=self.device)
+image_tensor = process_images([image], image_processor, model.config)
+image_tensor = [_image.to(dtype=torch.float16, device=device) for _image in image_tensor]
 
-conv_template = "llava_llama3" # Make sure you use correct chat template for different models
+conv_template = "llava_llama_3" # Make sure you use correct chat template for different models
 question = DEFAULT_IMAGE_TOKEN + "\nWhat is shown in this image?"
-conv = copy.deepcopy(conv_templates[self.conv_template])
+conv = copy.deepcopy(conv_templates[conv_template])
 conv.append_message(conv.roles[0], question)
 conv.append_message(conv.roles[1], None)
 prompt_question = conv.get_prompt()
-question_input.append(prompt_question)
 
-input_ids = tokenizer_image_token(prompt_question, tokenizer, IMAGE_TOKEN_INDEX, return_tensors="pt").to(self.device)
+input_ids = tokenizer_image_token(prompt_question, tokenizer, IMAGE_TOKEN_INDEX, return_tensors="pt").unsqueeze(0).to(device)
 image_sizes = [image.size]
 
 
-cont = self.model.generate(
+cont = model.generate(
  input_ids,
  images=image_tensor,
  image_sizes=image_sizes,
@@ -89,6 +96,7 @@ cont = self.model.generate(
  max_new_tokens=256,
 )
 text_outputs = tokenizer.batch_decode(cont, skip_special_tokens=True)
+print(text_outputs)
 
 ```