Initial commit

inference.py

@@ -0,0 +1,59 @@
+from llava.model.builder import load_pretrained_model
+from llava.mm_utils import get_model_name_from_path, process_images, tokenizer_image_token
+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
+from llava.actiontokenizer import ActionTokenizer
+from PIL import Image
+import requests
+import copy
+import torch
+
+import sys
+import warnings
+
+warnings.filterwarnings("ignore")
+pretrained = "/mnt/d/ckpt2/llava-openvla-qwen"
+model_name = "llava_qwen"
+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
+actiontokenizer = ActionTokenizer(tokenizer)
+
+print(model.model.embed_tokens.weight.shape)
+print(model.lm_head.weight.shape)
+
+model.eval()
+image_path = '/root/code/openvla-llava/LLaVA-Next/images/images/random_image.jpg'
+image = Image.open(image_path)
+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 = "qwen_1_5"  
+question = DEFAULT_IMAGE_TOKEN + "\n, the next position is:"
+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()
+
+input_ids = tokenizer_image_token(prompt_question, tokenizer, IMAGE_TOKEN_INDEX, return_tensors="pt").unsqueeze(0).to(device)
+image_sizes = [image.size]
+
+cont = model.generate(
+    input_ids,
+    images=image_tensor,
+    image_sizes=image_sizes,
+    do_sample=False,
+    temperature=0,
+    max_new_tokens=4096,
+)
+token_ids = cont.cpu().tolist()[0]
+decoded_text = []
+
+for token_id in token_ids:
+    if 151647 <= token_id <= 151903:
+        decoded_token = actiontokenizer.decode_token_ids_to_actions([token_id])
+    else:
+        decoded_token = tokenizer.decode([token_id])
+    decoded_text.append(decoded_token)
+final_text = ''.join(str(item) for item in decoded_text)
+print(final_text)