Update app.py

app.py

@@ -250,18 +250,44 @@ def process_audio_from_video(video_path):
 
 
 
-
-import torch
 import gradio as gr
-# Load model directly
+from huggingface_hub import InferenceClient
 from transformers import AutoTokenizer, AutoModelForCausalLM
+from huggingface_hub import InferenceClient
+
+client = InferenceClient("HuggingFaceH4/zephyr-7b-beta")
 
-tokenizer = AutoTokenizer.from_pretrained("TheBloke/Mistral-7B-Instruct-v0.1-GPTQ")
-model = AutoModelForCausalLM.from_pretrained("TheBloke/Mistral-7B-Instruct-v0.1-GPTQ")
 
+def respond(message, history: list[tuple[str, str]], system_message, max_tokens, temperature, top_p):
+    messages = [{"role": "system", "content": system_message}]
+    
+    # Format history with user and bot messages
+    for val in history:
+        if val[0]:
+            messages.append({"role": "user", "content": val[0]})
+        if val[1]:
+            messages.append({"role": "assistant", "content": val[1]})
+
+    messages.append({"role": "user", "content": message})
 
+    response = ""
+    
+    # Stream response from the model
+    for message in client.chat_completion(
+        messages,
+        max_tokens=max_tokens,
+        stream=True,
+        temperature=temperature,
+        top_p=top_p,
+    ):
+        token = message.choices[0].delta.content
+        response += token
+        yield response
+
+
+# Function to handle video processing and interaction
 def transcribe_and_predict_video(video, chat_history=[]):
-    # Process the video for emotions
+    # Process the video for emotions (use your own emotion detection functions)
     image_emotion = process_video(video)
     text_emotion, audio_emotion, user_input = process_audio_from_video(video)
     em = [image_emotion, text_emotion, audio_emotion]
@@ -272,12 +298,12 @@ def transcribe_and_predict_video(video, chat_history=[]):
     # Construct the prompt with emotion context and history
     prompt = f"""
     You are a helpful AI assistant. Respond like a human while considering the user's emotion.
-    
+
     User's Emotion: {em}
-    
+
     Conversation History:
     {history_text}
-    
+
     User ({em}): {user_input}
     Bot:"""
 
@@ -288,7 +314,7 @@ def transcribe_and_predict_video(video, chat_history=[]):
     output = model.generate(**inputs, max_length=512, temperature=0.7, top_p=0.9, do_sample=True)
     response = tokenizer.decode(output[0], skip_special_tokens=True).split("Bot:")[-1].strip()
 
-    # Store the current emotion for the user input (You can modify the emotion detection based on your needs)
+    # Store the current emotion for the user input (modify emotion detection as needed)
     emotion = detect_emotion(user_input)  # Assuming `detect_emotion` is a function that returns the user's emotion
 
     # Update the chat history with the current conversation and emotion
@@ -296,13 +322,22 @@ def transcribe_and_predict_video(video, chat_history=[]):
 
     return response, chat_history
 
-# Create Gradio interface
-iface = gr.Interface(fn=transcribe_and_predict_video, 
-                     inputs=gr.Video(), 
-                     outputs="text", 
-                     title="Multimodal Emotion Recognition from Video",
-                     description="Upload a video to get text, audio, and image emotion predictions.")
 
-iface.launch()
+# Gradio interface setup
+demo = gr.ChatInterface(
+    respond,
+    additional_inputs=[
+        gr.Textbox(value="You are a friendly Chatbot.", label="System message"),
+        gr.Slider(minimum=1, maximum=2048, value=512, step=1, label="Max new tokens"),
+        gr.Slider(minimum=0.1, maximum=4.0, value=0.7, step=0.1, label="Temperature"),
+        gr.Slider(minimum=0.1, maximum=1.0, value=0.95, step=0.05, label="Top-p (nucleus sampling)"),
+    ],
+)
+
+# Launch the Gradio interface
+if __name__ == "__main__":
+    demo.launch()
+
+