add cogagent

.gitignore

@@ -0,0 +1,10 @@
+en_core_web_sm-3.6.0/*
+daily_receive.*
+weekly_train_generate.*
+app[2-3].py
+test_gradio_client.py
+boxes.py
+*.tar.gz
+*.pyi
+__pycache__/
+examples/*grounding*

app.py

@@ -2,77 +2,38 @@
 
 import gradio as gr
 import os
-import re
-from PIL import Image
-import base64
-import time
+import json
+import requests
+from utils import is_chinese, process_image_without_resize, parse_response, templates_agent_cogagent, template_grounding_cogvlm, postprocess_text
+
+DESCRIPTION = '''<h2 style='text-align: center'> <a href="https://github.com/THUDM/CogVLM"> CogVLM & CogAgent Chat Demo</a> </h2>'''
 
-DESCRIPTION = '''# <a href="https://github.com/THUDM/CogVLM">VisualGLM</a>'''
+NOTES = 'This app is adapted from <a href="https://github.com/THUDM/CogVLM">https://github.com/THUDM/CogVLM</a>. It would be recommended to check out the repo if you want to see the detail of our model.\n\n该demo仅作为测试使用，不支持批量请求。如有大批量需求，欢迎联系[智谱AI](mailto:business@zhipuai.cn)。\n\n请注意CoogVLM-17B目前仅支持英文。'
 
 MAINTENANCE_NOTICE1 = 'Hint 1: If the app report "Something went wrong, connection error out", please turn off your proxy and retry.<br>Hint 2: If you upload a large size of image like 10MB, it may take some time to upload and process. Please be patient and wait.'
 
-GROUNDING_NOTICE = 'Hint: When you check "Grounding", please use the <a href="https://github.com/THUDM/CogVLM/blob/main/utils/template.py#L344">corresponding prompt</a> or the examples below.'
+GROUNDING_NOTICE = 'Hint: When you check "Grounding", please use the <a href="https://github.com/THUDM/CogVLM/blob/main/utils/utils/template.py#L344">corresponding prompt</a> or the examples below.'
 
+AGENT_NOTICE = 'Hint: When you check "CogAgent", please use the <a href="https://github.com/THUDM/CogVLM/blob/main/utils/utils/template.py#L761C1-L761C17">corresponding prompt</a> or the examples below.'
 
-NOTES = 'This app is adapted from <a href="https://github.com/THUDM/CogVLM">https://github.com/THUDM/CogVLM</a>. It would be recommended to check out the repo if you want to see the detail of our model.'
-
-import json
-import requests
-import base64
-import hashlib
-from utils import parse_response
 
 default_chatbox = [("", "Hi, What do you want to know about this image?")]
 
 URL = os.environ.get("URL")
 
-def process_image(image_prompt):
-    image = Image.open(image_prompt)
-    print(f"height:{image.height}, width:{image.width}")
-    resized_image = image.resize((224, 224), )
-    timestamp = int(time.time())
-    file_ext = os.path.splitext(image_prompt)[1]
-    filename = f"examples/{timestamp}{file_ext}"
-    resized_image.save(filename)
-    print(f"temporal filename {filename}")
-    with open(filename, "rb") as image_file:
-        bytes = base64.b64encode(image_file.read())
-        encoded_img = str(bytes, encoding='utf-8')
-        image_hash = hashlib.sha256(bytes).hexdigest()
-        os.remove(filename)
-        return encoded_img, image_hash
-
-
-def process_image_without_resize(image_prompt):
-    image = Image.open(image_prompt)
-    print(f"height:{image.height}, width:{image.width}")
-    timestamp = int(time.time())
-    file_ext = os.path.splitext(image_prompt)[1]
-    filename = f"examples/{timestamp}{file_ext}"
-    filename_grounding = f"examples/{timestamp}_grounding{file_ext}"
-    image.save(filename)
-    print(f"temporal filename {filename}")
-    with open(filename, "rb") as image_file:
-        bytes = base64.b64encode(image_file.read())
-        encoded_img = str(bytes, encoding='utf-8')
-        image_hash = hashlib.sha256(bytes).hexdigest()
-        os.remove(filename)
-        return image, encoded_img, image_hash, filename_grounding
-
-
-def is_chinese(text):
-    zh_pattern = re.compile(u'[\u4e00-\u9fa5]+')
-    return zh_pattern.search(text)
-
 
 def post(
         input_text,
         temperature,
         top_p,
+        top_k,
         image_prompt,
         result_previous,
         hidden_image,
-        grounding    
+        grounding,
+        cogagent,
+        grounding_template,
+        agent_template    
         ):
     result_text = [(ele[0], ele[1]) for ele in result_previous]
     for i in range(len(result_text)-1, -1, -1):
@@ -107,16 +68,35 @@ def post(
             result_text = []        
         hidden_image = image_hash        
     else:
-        encoded_img = None
+        encoded_img = None 
+
+    model_use = "vlm_chat"
+    if not cogagent and grounding:
+        model_use = "vlm_grounding"
+        if grounding_template:
+            input_text = postprocess_text(grounding_template, input_text)
+    elif cogagent:
+        model_use = "agent_chat"
+        if agent_template:
+            input_text = postprocess_text(agent_template, input_text)
+
+    prompt = input_text
 
-    print('request chat model...' if not grounding else 'request grounding model...')
+    if grounding:
+        prompt += "(with grounding)"   
+
+    print(f'request {model_use} model... with prompt {prompt}, grounding_template {grounding_template}, agent_template {agent_template}')
     data = json.dumps({
-        'text': input_text,
+        'model_use': model_use,
+        'is_grounding': grounding,
+        'text': prompt,
+        'history': result_text,
         'image': encoded_img,
         'temperature': temperature,
         'top_p': top_p,
-        'history': result_text,
-        'is_grounding': grounding
+        'top_k': top_k,
+        'do_sample': True,
+        'max_new_tokens': 2048
     })
     try:
         response = requests.request("POST", URL, headers=headers, data=data, timeout=(60, 100)).json()
@@ -161,6 +141,9 @@ def main():
 
     with gr.Blocks(css='style.css') as demo:
 
+        gr.Markdown(DESCRIPTION)
+        gr.Markdown(NOTES)
+
         with gr.Row():
             with gr.Column(scale=4.5):
                 with gr.Group():
@@ -172,28 +155,33 @@ def main():
                     image_prompt = gr.Image(type="filepath", label="Image Prompt", value=None)
                 with gr.Row():
                     grounding = gr.Checkbox(label="Grounding")
+                    cogagent = gr.Checkbox(label="CogAgent")
                 with gr.Row():
-                    grounding_notice = gr.Markdown(GROUNDING_NOTICE)
+                    # grounding_notice = gr.Markdown(GROUNDING_NOTICE)
+                    grounding_template = gr.Dropdown(choices=template_grounding_cogvlm, label="Grounding Template", value=template_grounding_cogvlm[0])
+                    # agent_notice = gr.Markdown(AGENT_NOTICE)
+                    agent_template = gr.Dropdown(choices=templates_agent_cogagent, label="Agent Template", value=templates_agent_cogagent[0])
 
                 with gr.Row():
-                    temperature = gr.Slider(maximum=1, value=0.8, minimum=0, label='Temperature')
-                    top_p = gr.Slider(maximum=1, value=0.4, minimum=0, label='Top P')
+                    temperature = gr.Slider(maximum=1, value=0.9, minimum=0, label='Temperature')
+                    top_p = gr.Slider(maximum=1, value=0.8, minimum=0, label='Top P')
+                    top_k = gr.Slider(maximum=50, value=5, minimum=1, step=1, label='Top K')
+
             with gr.Column(scale=5.5):
-                result_text = gr.components.Chatbot(label='Multi-round conversation History', value=[("", "Hi, What do you want to know about this image?")]).style(height=550)
+                result_text = gr.components.Chatbot(label='Multi-round conversation History', value=[("", "Hi, What do you want to know about this image?")], height=550)
                 hidden_image_hash = gr.Textbox(visible=False)
 
-        gr_examples = gr.Examples(examples=[[example["text"], example["image"]] for example in examples], 
-                                  inputs=[input_text, image_prompt],
+        gr_examples = gr.Examples(examples=[[example["text"], example["image"], example["grounding"], example["cogagent"]] for example in examples], 
+                                  inputs=[input_text, image_prompt, grounding, cogagent],
                                   label="Example Inputs (Click to insert an examplet into the input box)",
                                   examples_per_page=6)
 
         gr.Markdown(MAINTENANCE_NOTICE1)
-        gr.Markdown(NOTES)
 
         print(gr.__version__)
-        run_button.click(fn=post,inputs=[input_text, temperature, top_p, image_prompt, result_text, hidden_image_hash, grounding],
+        run_button.click(fn=post,inputs=[input_text, temperature, top_p, top_k, image_prompt, result_text, hidden_image_hash, grounding, cogagent, grounding_template, agent_template],
                          outputs=[input_text, result_text, hidden_image_hash])
-        input_text.submit(fn=post,inputs=[input_text, temperature, top_p, image_prompt, result_text, hidden_image_hash, grounding],
+        input_text.submit(fn=post,inputs=[input_text, temperature, top_p, top_k, image_prompt, result_text, hidden_image_hash, grounding, cogagent, grounding_template, agent_template],
                          outputs=[input_text, result_text, hidden_image_hash])
         clear_button.click(fn=clear_fn, inputs=clear_button, outputs=[input_text, result_text, image_prompt])
         image_prompt.upload(fn=clear_fn2, inputs=clear_button, outputs=[result_text])
@@ -202,7 +190,7 @@ def main():
         print(gr.__version__)
 
     demo.queue(concurrency_count=10)
-    demo.launch()
+    demo.launch(server_port=7862)
 
 if __name__ == '__main__':
     main()

examples/example_inputs.jsonl

@@ -1,6 +1,6 @@
-{"id":1, "text": "Describe this image", "image": "examples/1.png"}
-{"id":2, "text": "What is written in the image?", "image": "examples/2.jpg"}
-{"id":3, "text": "How many houses are there in this cartoon?", "image": "examples/3.jpg"}
-{"id":4, "text": "Can you provide a description of the image and include the coordinates [[x0,y0,x1,y1]] for each mentioned object?", "image": "examples/4.png"}
-{"id":5, "text": "Where is the tree closer to the sun?", "image": "examples/5.jpg"}
-{"id":6, "text": "What color are the clothes of the girl whose hands are holding flowers? Let's think step by step", "image": "examples/6.jpg"}
+{"id":1, "text": "Describe this image", "image": "examples/1.png", "grounding": false, "cogagent": false}
+{"id":2, "text": "What is written in the image?", "image": "examples/2.jpg", "grounding": false, "cogagent": false}
+{"id":3, "text": "the tree closer to the sun", "image": "examples/3.jpg", "grounding": true, "cogagent": false}
+{"id":4, "text": "What color are the clothes of the girl whose hands are holding flowers? Let's think step by step", "image": "examples/4.jpg", "grounding": true, "cogagent": false}
+{"id":5, "text": "search CogVLM", "image": "examples/5.jpeg", "grounding": true, "cogagent": true}
+{"id":6, "text": "Insert a new slide named 'In-context learning: Details' with a Two Content layout after the current slide.", "image": "examples/6.jpeg", "grounding": false, "cogagent": true}

utils.py

@@ -3,9 +3,38 @@ from PIL import Image, ImageDraw, ImageFont
 import matplotlib.font_manager
 import spacy
 import re
+import base64
+import time
+import re
+from PIL import Image
+import base64
+import hashlib
+import os
 
 nlp = spacy.load("en_core_web_sm-3.6.0")
 
+def process_image_without_resize(image_prompt):
+    image = Image.open(image_prompt)
+    print(f"height:{image.height}, width:{image.width}")
+    timestamp = time.time()
+    file_ext = os.path.splitext(image_prompt)[1]
+    filename = f"examples/{timestamp}{file_ext}"
+    filename_grounding = f"examples/{timestamp}_grounding{file_ext}"
+    image.save(filename)
+    print(f"temporal filename {filename}")
+    with open(filename, "rb") as image_file:
+        bytes = base64.b64encode(image_file.read())
+        encoded_img = str(bytes, encoding='utf-8')
+        image_hash = hashlib.sha256(bytes).hexdigest()
+        os.remove(filename)
+        return image, encoded_img, image_hash, filename_grounding
+
+
+def is_chinese(text):
+    zh_pattern = re.compile(u'[\u4e00-\u9fa5]+')
+    return zh_pattern.search(text)
+
+
 def draw_boxes(image, boxes, texts, output_fn='output.png'):
     box_width = 5
     color_palette = sns.color_palette("husl", len(boxes))
@@ -83,4 +112,312 @@ def parse_response(img, response, output_fn='output.png'):
         boxes = []
     else:
         texts, boxes = zip(*dic.items())
-    draw_boxes(new_img, boxes, texts, output_fn=output_fn)
+    draw_boxes(new_img, boxes, texts, output_fn=output_fn)
+
+def postprocess_text(template, text):
+    quoted_text = f'"{text.strip()}"'
+    return template.replace("<TASK>", quoted_text).strip() if template != "" else text.strip()
+
+    
+# The templates is for CogAgent_Agent Template
+templates_agent_cogagent = [
+    "Can you advise me on how to <TASK>?",
+    "I'm looking for guidance on how to <TASK>.",
+    "What steps do I need to take to <TASK>?",
+    "Could you provide instructions for <TASK>?",
+    "I'm wondering what the process is for <TASK>.",
+    "How can I go about <TASK>?",
+    "I need assistance with planning to <TASK>.",
+    "Do you have any recommendations for <TASK>?",
+    "Please share some tips for <TASK>.",
+    "I'd like to know the best way to <TASK>.",
+    "What's the most effective way to <TASK>?",
+    "I'm seeking advice on accomplishing <TASK>.",
+    "Could you guide me through the steps to <TASK>?",
+    "I'm unsure how to start with <TASK>.",
+    "Is there a strategy for successfully <TASK>?",
+    "What's the proper procedure for <TASK>?",
+    "How should I prepare for <TASK>?",
+    "I'm not sure where to begin with <TASK>.",
+    "I need some insights on <TASK>.",
+    "Can you explain how to tackle <TASK>?",
+    "I'm interested in the process of <TASK>.",
+    "Could you enlighten me on <TASK>?",
+    "What are the recommended steps for <TASK>?",
+    "Is there a preferred method for <TASK>?",
+    "I'd appreciate your advice on <TASK>.",
+    "Can you shed light on <TASK>?",
+    "What would be the best approach to <TASK>?",
+    "How do I get started with <TASK>?",
+    "I'm inquiring about the procedure for <TASK>.",
+    "Could you share your expertise on <TASK>?",
+    "I'd like some guidance on <TASK>.",
+    "What's your recommendation for <TASK>?",
+    "I'm seeking your input on how to <TASK>.",
+    "Can you provide some insights into <TASK>?",
+    "How can I successfully accomplish <TASK>?",
+    "What steps are involved in <TASK>?",
+    "I'm curious about the best way to <TASK>.",
+    "Could you show me the ropes for <TASK>?",
+    "I need to know how to go about <TASK>.",
+    "What are the essential steps for <TASK>?",
+    "Is there a specific method for <TASK>?",
+    "I'd like to get some advice on <TASK>.",
+    "Can you explain the process of <TASK>?",
+    "I'm looking for guidance on how to approach <TASK>.",
+    "What's the proper way to handle <TASK>?",
+    "How should I proceed with <TASK>?",
+    "I'm interested in your expertise on <TASK>.",
+    "Could you walk me through the steps for <TASK>?",
+    "I'm not sure where to begin when it comes to <TASK>.",
+    "What should I prioritize when doing <TASK>?",
+    "How can I ensure success with <TASK>?",
+    "I'd appreciate some tips on <TASK>.",
+    "Can you provide a roadmap for <TASK>?",
+    "What's the recommended course of action for <TASK>?",
+    "I'm seeking your guidance on <TASK>.",
+    "Could you offer some suggestions for <TASK>?",
+    "I'd like to know the steps to take for <TASK>.",
+    "What's the most effective way to achieve <TASK>?",
+    "How can I make the most of <TASK>?",
+    "I'm wondering about the best approach to <TASK>.",
+    "Can you share your insights on <TASK>?",
+    "What steps should I follow to complete <TASK>?",
+    "I'm looking for advice on <TASK>.",
+    "What's the strategy for successfully completing <TASK>?",
+    "How should I prepare myself for <TASK>?",
+    "I'm not sure where to start with <TASK>.",
+    "What's the procedure for <TASK>?",
+    "Could you provide some guidance on <TASK>?",
+    "I'd like to get some tips on how to <TASK>.",
+    "Can you explain how to tackle <TASK> step by step?",
+    "I'm interested in understanding the process of <TASK>.",
+    "What are the key steps to <TASK>?",
+    "Is there a specific method that works for <TASK>?",
+    "I'd appreciate your advice on successfully completing <TASK>.",
+    "Can you shed light on the best way to <TASK>?",
+    "What would you recommend as the first step to <TASK>?",
+    "How do I initiate <TASK>?",
+    "I'm inquiring about the recommended steps for <TASK>.",
+    "Could you share some insights into <TASK>?",
+    "I'm seeking your expertise on <TASK>.",
+    "What's your recommended approach for <TASK>?",
+    "I'd like some guidance on where to start with <TASK>.",
+    "Can you provide recommendations for <TASK>?",
+    "What's your advice for someone looking to <TASK>?",
+    "I'm seeking your input on the process of <TASK>.",
+    "How can I achieve success with <TASK>?",
+    "What's the best way to navigate <TASK>?",
+    "I'm curious about the steps required for <TASK>.",
+    "Could you show me the proper way to <TASK>?",
+    "I need to know the necessary steps for <TASK>.",
+    "What's the most efficient method for <TASK>?",
+    "I'd appreciate your guidance on <TASK>.",
+    "Can you explain the steps involved in <TASK>?",
+    "I'm looking for recommendations on how to approach <TASK>.",
+    "What's the right way to handle <TASK>?",
+    "How should I manage <TASK>?",
+    "I'm interested in your insights on <TASK>.",
+    "Could you provide a step-by-step guide for <TASK>?",
+    "I'm not sure how to start when it comes to <TASK>.",
+    "What are the key factors to consider for <TASK>?",
+    "How can I ensure a successful outcome with <TASK>?",
+    "I'd like some tips and tricks for <TASK>.",
+    "Can you offer a roadmap for accomplishing <TASK>?",
+    "What's the preferred course of action for <TASK>?",
+    "I'm seeking your expert advice on <TASK>.",
+    "Could you suggest some best practices for <TASK>?",
+    "I'd like to understand the necessary steps to complete <TASK>.",
+    "What's the most effective strategy for <TASK>?",
+]
+
+template_grounding_cogvlm = [
+    "Where is <TASK>?",
+    "Where is <TASK> in the image?",
+    "Where is <TASK>? answer in [[x0,y0,x1,y1]] format.",
+    "Can you point out <TASK> in the image and provide the bounding boxes of its location?",
+    "Help me to locate <TASK> in and give me its bounding boxes, please.",
+    "In the given, could you find and tell me the bounding boxes of <TASK>?",
+    "Guide me to the location of <TASK> within the image by providing its bounding boxes.",
+    "I'd like to know the exact bounding boxes of <TASK> in the photo.",
+    "Would you kindly provide the bounding boxes of <TASK> located in the picture?",
+    "Can you find <TASK> in and give me the bounding boxes of where it is located?",
+    "I'm trying to locate <TASK> in. Can you determine its bounding boxes for me?",
+    "What are the bounding boxes of <TASK> in the image?",
+    "Can you disclose the position of <TASK> in the photograph by stating its bounding boxes?",
+    "In, could you let me know the location of <TASK> in the form of bounding boxes?",
+    "I need the bounding boxes of <TASK> in, can you please assist me with that?",
+    "Where in is <TASK> located? Provide me with its bounding boxes, please.",
+    "May I have the bounding boxes of <TASK>?",
+    "In the photograph, could you pinpoint the location of <TASK> and tell me its bounding boxes?",
+    "Can you please search and find <TASK> in, then let me know its bounding boxes?",
+    "Please, point out the position of <TASK> in the image by giving its bounding boxes.",
+    "What are the exact bounding boxes of <TASK> in the provided picture?",
+    "Detect the location of <TASK> in and share the bounding boxes with me, please.",
+    "In the picture, I'd like you to locate <TASK> and provide its coordinates.",
+    "Please indicate the location of <TASK> in the photo by giving bounding boxes.",
+    "Find <TASK> in and share its coordinates with me.",
+    "Could you please help me find the bounding boxes of <TASK> in the image?",
+    "I am looking for the position of <TASK> in. Can you provide its bounding boxes?",
+    "In the image, can you locate <TASK> and let me know its coordinates?",
+    "I'd appreciate if you could find and tell me the bounding boxes of <TASK>.",
+    "In, I need the bounding box bounding boxes of <TASK>.",
+    "Point me to the location of <TASK> in the picture by providing its bounding boxes.",
+    "Could you trace <TASK> in and tell me its bounding boxes?",
+    "Can you assist me in locating <TASK> in, and then provide its bounding boxes?",
+    "I'm curious, what are the bounding boxes of <TASK> in the photo?",
+    "Kindly share the bounding boxes of <TASK> located in the image.",
+    "I would like to find <TASK> in. Can you give me its bounding boxes?",
+    "Can you spot <TASK> in and disclose its bounding boxes to me?",
+    "Please, reveal the location of <TASK> in the provided photograph as coordinates.",
+    "Help me locate and determine the bounding boxes of <TASK>.",
+    "I request the bounding boxes of <TASK> in the image.",
+    "In the given, can you find <TASK> and tell me its bounding boxes?",
+    "I need to know the position of <TASK> in as bounding boxes.",
+    "Locate <TASK> in and provide its bounding boxes, please.",
+    "Assist me in finding <TASK> in the photo and provide the bounding box bounding boxes.",
+    "In, can you guide me to the location of <TASK> by providing bounding boxes?",
+    "I'd like the bounding boxes of <TASK> as it appears in the image.",
+    "What location does <TASK> hold in the picture? Inform me of its bounding boxes.",
+    "Identify the position of <TASK> in and share its bounding boxes.",
+    "I'd like to request the bounding boxes of <TASK> within the photo.",
+    "How can I locate <TASK> in the image? Please provide the bounding boxes.",
+    "I am interested in knowing the bounding boxes of <TASK> in the picture.",
+    "Assist me in locating the position of <TASK> in the photograph and its bounding box bounding boxes.",
+    "In the image, I need to find <TASK> and know its bounding boxes. Can you please help?"
+    "Can you give me a description of the region <TASK> in image?",
+    "In the provided image, would you mind describing the selected area <TASK>?",
+    "I need details about the area <TASK> located within image.",
+    "Could you please share some information on the region <TASK> in this photograph?",
+    "Describe what's happening within the coordinates <TASK> of the given image.",
+    "What can you tell me about the selected region <TASK> in the photo?",
+    "Please, can you help me understand what's inside the region <TASK> in image?",
+    "Give me a comprehensive description of the specified area <TASK> in the picture.",
+    "I'm curious about the area <TASK> in the following image. Can you describe it?",
+    "Please elaborate on the area with the coordinates <TASK> in the visual.",
+    "In the displayed image, help me understand the region defined by <TASK>.",
+    "Regarding the image, what's going on in the section <TASK>?",
+    "In the given photograph, can you explain the area with coordinates <TASK>?",
+    "Kindly describe what I should be seeing in the area <TASK> of image.",
+    "Within the input image, what can be found in the region defined by <TASK>?",
+    "Tell me what you see within the designated area <TASK> in the picture.",
+    "Please detail the contents of the chosen region <TASK> in the visual input.",
+    "What's inside the area <TASK> of the provided graphic?",
+    "I'd like some information about the specific region <TASK> in the image.",
+    "Help me understand the details within the area <TASK> in photograph.",
+    "Can you break down the region <TASK> in the image for me?",
+    "What is taking place within the specified area <TASK> in this capture?",
+    "Care to elaborate on the targeted area <TASK> in the visual illustration?",
+    "What insights can you provide about the area <TASK> in the selected picture?",
+    "What does the area <TASK> within the given visual contain?",
+    "Analyze and describe the region <TASK> in the included photo.",
+    "Please provide details for the area marked as <TASK> in this photographic.",
+    "For the image, can you assess and describe what's happening at <TASK>?",
+    "Fill me in about the selected portion <TASK> within the presented image.",
+    "In the image, elaborate on the details found within the section <TASK>.",
+    "Please interpret and describe the area <TASK> inside the given picture.",
+    "What information can you give me about the coordinates <TASK> in image?",
+    "Regarding the coordinates <TASK> in image, can you provide a description?",
+    "In the photo, can you delve into the details of the region <TASK>?",
+    "Please provide insights on the specified area <TASK> within the graphic.",
+    "Detail the chosen region <TASK> in the depicted scene.",
+    "Can you discuss the entities within the region <TASK> of image?",
+    "I'd appreciate a breakdown of the area <TASK> in the displayed image.",
+    "What's the story in the section <TASK> of the included visual?",
+    "Please enlighten me about the region <TASK> in the given photo.",
+    "Offer a thorough description of the area <TASK> within the illustration.",
+    "What can you share about the area <TASK> in the presented image?",
+    "Help me grasp the context of the region <TASK> within image.",
+    "Kindly give an overview of the section <TASK> in photo.",
+    "What details can you provide about the region <TASK> in the snapshot?",
+    "Can you divulge the contents of the area <TASK> within the given image?",
+    "In the submitted image, please give a synopsis of the area <TASK>.",
+    "In the image, please describe the bounding box <TASK>.",
+    "Please describe the region <TASK> in the picture.",
+    "Describe the bbox <TASK> in the provided photo.",
+    "What can you tell me about the area <TASK> within the image?",
+    "Could you give me a description of the rectangular region <TASK> found in?",
+    "In, what elements can be found within the coordinates <TASK>?",
+    "Please provide details for the area within the bounding box <TASK> in.",
+    "Can you generate a description for the selected region <TASK> in the image?",
+    "Kindly describe the objects or scenery in the bounding box <TASK> within.",
+    "What details can you provide for the rectangle defined by the coordinates <TASK> in?",
+    "In relation to the picture, please describe the content of the area marked by <TASK>.",
+    "I'd like to know more about the area <TASK> in the given image. Can you describe it?",
+    "Can you help me by describing the part of that lies within the bounding box <TASK>?",
+    "What's happening in the section of the photo enclosed by the coordinates <TASK>?",
+    "Describe the image content present in the specified rectangular area <TASK> of.",
+    "Please provide information about the area within the bounding box <TASK> in the picture.",
+    "Could you offer a description of the contents in the selected area <TASK> of the image?",
+    "I'm curious about the area <TASK> in. Can you provide a description of it?",
+    "What can be observed in the rectangular region <TASK> in the photograph?",
+    "Please explain what is contained in the portion of defined by the box <TASK>.",
+    "In the photograph, can you describe the objects or scenery enclosed by <TASK>?",
+    "Can you give a brief explanation of the specified area <TASK> in the image?",
+    "What does the area <TASK> look like in the context of the image?",
+    "Could you please describe the contents of the bounding box <TASK> in the given image?",
+    "I would like to know more about the rectangular region <TASK> within the picture. Can you describe it?",
+    "Please tell me about the area <TASK> in the image. What does it contain?",
+    "Help me understand what's happening in the selected bounding box <TASK> within.",
+    "Can you provide a description of the area <TASK> in the image?",
+    "What sort of things can be seen in the region <TASK> of the photo?",
+    "Describe what can be found within the bounds of <TASK> in the image.",
+    "In, can you paint a picture of the area enclosed by coordinates <TASK>?",
+    "Please provide a detailed account of the area covered by the bounding box <TASK> in.",
+    "Give me a vivid description of what's happening in the area <TASK> within the snapshot.",
+    "In the image, what do you observe within the rectangular box defined by the coordinates <TASK>?",
+    "Could you give me a breakdown of the content in the specified area <TASK> of the picture?",
+    "Please elucidate the area<TASK> of the image.",
+    "I'd appreciate it if you could describe the portion of that lies within the rectangle <TASK>.",
+    "Can you share some insights about the rectangular region <TASK> in the image?",
+    "Help me visualize the section of the photo enclosed by the bounding box <TASK>.",
+    "Would you kindly provide a description for the content within the rectangular area <TASK> of?",
+    "In, can you tell me more about the area specified by the bounding box <TASK>?",
+    "Please describe what can be seen in the rectangular region <TASK> of the image.",
+    "Can you analyze the content of the area <TASK> within the photograph?",
+    "In the provided image, please explain the content within the region <TASK>.",
+    "I'm interested in the selected rectangle <TASK> in. Can you tell me more about it?",
+    "Explain what can be found in the bounding box <TASK> in the context of the image.",
+    "Kindly share your observations about the rectangular region <TASK> within.",
+    "I'd like a thorough description of the area <TASK> in the image.",
+    "Could you please provide a description of the rectangular area <TASK> in?",
+    "Please describe the section of the picture defined by the bbox <TASK>.",
+    "Tell me more about the scenery or objects within the rectangular region <TASK> in.",
+    "Would you kindly describe the content of the area enclosed by <TASK> in the image?",
+    "Help me understand the objects or scenery within the bounding box <TASK> in the image.",
+    "I would like to know about the section of the image enclosed by the rectangle <TASK>. Can you describe it?",
+    "Describe the selected rectangular area <TASK> in the photo.",
+    "Tell me about the region <TASK> of the image.",
+    "I request a description of the area <TASK> in the picture.",
+    "Can you elaborate on the content of the bounding box <TASK> in?",
+    "Please share details about the rectangular region <TASK> within the image.",
+    "What can I find in the bbox <TASK> of the provided image?",
+    "In the image, could you provide a description for the coordinates <TASK>?",
+    "Could you tell me more about the area <TASK> in the snapshot?",
+    "Fill me in on the details of the rectangular box <TASK> within the image.",
+    "What's going on in the section of contained within the bounding box <TASK>?",
+    "I would like a description of the content within the bbox <TASK> in.",
+    "Please enlighten me about the area <TASK> in the photograph.",
+    "Can you give me a visual rundown of the area <TASK> in?",
+    "Describe the visual elements within the selected area <TASK> of the image.",
+    "Tell me what you see in the area <TASK> within the context of the image.",
+    "Explain the content within the rectangular region <TASK> of the image.",
+    "I'd like some information about the bounding box <TASK> in the photo.",
+    "What is happening within the rectangle defined by coordinates <TASK> in the image?",
+    "Please describe the content within the area <TASK> displayed in the image.",
+    "What can be seen in the bounding box <TASK> in the context of the provided image?",
+    "Share some details about the objects or environment within the bounding box <TASK> in.",
+    "Please describe the area <TASK> in the image for me.",
+    "Can you generate a description of the contents within the selected region <TASK> in?",
+    "What objects or scenery can be found in the area <TASK> in the image?",
+    "Please tell me more about the rectangular section <TASK> in the photo.",
+    "Could you describe the content of the bbox <TASK> in the image?",
+    "What does the selected region <TASK> in the image encompass?",
+    "I am interested in the region <TASK> of the image; please describe it.",
+    "Can you provide some context for the area <TASK> within the picture?",
+    "Please give me some details about the rectangle <TASK> in the image.",
+    "In the photo, what can you see within the region defined by the bounding box <TASK>?",
+    "I would like a detailed description of the portion of enclosed by the bbox <TASK>.",
+    "Please help me understand the content present within the rectangle <TASK> in.",
+    "Would you mind describing the rectangular area <TASK> in the provided image?"
+]