Create app.py

app.py

@@ -0,0 +1,33 @@
+import gradio as gr
+import torch
+from transformers import AutoModel, AutoTokenizer
+from huggingface_hub import hf_hub_download
+import spaces
+
+# Download the model and tokenizer
+model_name = 'internlm/internlm-xcomposer2d5-7b'
+model = AutoModel.from_pretrained(model_name, torch_dtype=torch.bfloat16, trust_remote_code=True)
+tokenizer = AutoTokenizer.from_pretrained(model_name, trust_remote_code=True)
+model.tokenizer = tokenizer
+
+@spaces.GPU
+def process_video(video_path, query):
+    torch.set_grad_enabled(False)
+    
+    with torch.autocast(device_type='cuda', dtype=torch.float16):
+        response, history = model.chat(tokenizer, query, [video_path], do_sample=False, num_beams=3, use_meta=True)
+    
+    return response
+
+iface = gr.Interface(
+    fn=process_video,
+    inputs=[
+        gr.Video(label="Upload Video"),
+        gr.Textbox(label="Enter your query")
+    ],
+    outputs=gr.Textbox(label="Response"),
+    title="Video Analysis with InternLM-XComposer",
+    description="Upload a video and ask a question about it."
+)
+
+iface.launch(debug=True)