Update app.py

app.py

@@ -6,16 +6,26 @@ import json
 import os
 
 
+os.system("pip install matplotlib==3.10.6")
+
+os.system("pip install Pillow==11.3.0")
+os.system("pip install qwen_vl_utils==0.0.14")
+os.system("pip install transformers==4.51.3")
+# os.system("pip install vllm==0.8.2")
+os.system("pip install accelerate==1.10.1")
+
+os.system("pip install gradio==4.44.1")
+os.system("pip install gradio_image_prompter==0.1.0")
+# os.system("pip install pydantic==2.10.6")
+os.system("pip install torch==2.4.0 torchvision==0.18.0 --index-url https://download.pytorch.org/whl/cu124")
 
 import subprocess
 subprocess.run('pip install flash-attn==2.7.4.post1 --no-build-isolation', env={'FLASH_ATTENTION_SKIP_CUDA_BUILD': "TRUE"}, shell=True)
-os.system("pip install transformers==4.51.1")
-os.system("pip install numpy==1.26.4")
-os.system("pip install vllm==0.8.2")
-os.system("pip install torch==2.4.0 torchvision==0.18.0 --index-url https://download.pytorch.org/whl/cu124")
 
 import sys
 import threading
+# os.system("pip install torch==2.5.1 torchvision==0.20.1 torchaudio==2.5.1 --index-url https://download.pytorch.org/whl/cu124")
+os.system("pip install --no-cache-dir --force-reinstall 'numpy==1.26.4'")
 import re
 from typing import Any, Dict, List
 
@@ -29,7 +39,7 @@ from rex_omni.tasks import KEYPOINT_CONFIGS, TASK_CONFIGS, get_task_config
 
 
 def parse_args():
-    parser = argparse.ArgumentParser(description="Rex-Omni Gradio Demo")
+    parser = argparse.ArgumentParser(description="Rex Omni Gradio Demo")
     parser.add_argument(
         "--model_path",
         default="IDEA-Research/Rex-Omni",
@@ -45,7 +55,7 @@ def parse_args():
     parser.add_argument("--temperature", type=float, default=0.0)
     parser.add_argument("--top_p", type=float, default=0.05)
     parser.add_argument("--top_k", type=int, default=1)
-    parser.add_argument("--max_tokens", type=int, default=4096)
+    parser.add_argument("--max_tokens", type=int, default=2048)
     parser.add_argument("--repetition_penalty", type=float, default=1.05)
     parser.add_argument("--min_pixels", type=int, default=16 * 28 * 28)
     parser.add_argument("--max_pixels", type=int, default=2560 * 28 * 28)
@@ -942,4 +952,4 @@ if __name__ == "__main__":
         server_port=args.server_port,
         share=True,
         debug=True,
-    )
+    )