Update app.py

app.py

@@ -1,19 +1,71 @@
 import gradio as gr
-from huggingface_hub import InferenceClient
 import os
 import pandas as pd
+import requests
+import json
 from typing import List, Tuple
 
-# LLM Models Definition
-LLM_MODELS = {
-    "Cohere c4ai-crp-08-2024": "CohereForAI/c4ai-command-r-plus-08-2024",  # Default
-    "Meta Llama3.3-70B": "meta-llama/Llama-3.3-70B-Instruct",    
-    "Mistral Nemo 2407": "mistralai/Mistral-Nemo-Instruct-2407",
-    "Alibaba Qwen QwQ-32B": "Qwen/QwQ-32B-Preview"
-}
-
-def get_client(model_name):
-    return InferenceClient(LLM_MODELS[model_name], token=os.getenv("HF_TOKEN"))
+class OllamaClient:
+    def __init__(self, model_name: str = "mistral-nemo", base_url: str = "http://localhost:11434"):
+        self.model_name = model_name
+        self.base_url = base_url
+    
+    def chat_completion(self, messages, max_tokens=4000, stream=True, temperature=0.7, top_p=0.9):
+        # Convert messages to Ollama format
+        ollama_messages = []
+        for msg in messages:
+            if msg["role"] == "system":
+                ollama_messages.append({"role": "system", "content": msg["content"]})
+            elif msg["role"] in ["user", "assistant"]:
+                ollama_messages.append({"role": msg["role"], "content": msg["content"]})
+        
+        # Prepare the request data
+        data = {
+            "model": self.model_name,
+            "messages": ollama_messages,
+            "options": {
+                "temperature": temperature,
+                "top_p": top_p,
+                "num_predict": max_tokens
+            },
+            "stream": stream
+        }
+        
+        # Make the request to Ollama API
+        response = requests.post(
+            f"{self.base_url}/api/chat",
+            json=data,
+            stream=stream
+        )
+        
+        if response.status_code != 200:
+            raise Exception(f"Ollama API error: {response.text}")
+        
+        if stream:
+            for line in response.iter_lines():
+                if line:
+                    decoded_line = line.decode('utf-8')
+                    try:
+                        chunk = json.loads(decoded_line)
+                        if "message" in chunk:
+                            yield {
+                                "choices": [{
+                                    "delta": {
+                                        "content": chunk["message"]["content"]
+                                    }
+                                }]
+                            }
+                    except json.JSONDecodeError:
+                        continue
+        else:
+            result = response.json()
+            yield {
+                "choices": [{
+                    "delta": {
+                        "content": result["message"]["content"]
+                    }
+                }]
+            }
 
 def analyze_file_content(content, file_type):
     """Analyze file content and return structural summary"""
@@ -94,7 +146,7 @@ def format_history(history):
             formatted_history.append({"role": "assistant", "content": assistant_msg})
     return formatted_history
 
-def chat(message, history, uploaded_file, model_name, system_message="", max_tokens=4000, temperature=0.7, top_p=0.9):
+def chat(message, history, uploaded_file, system_message="", max_tokens=4000, temperature=0.7, top_p=0.9):
     system_prefix = """You are a file analysis expert. Analyze the uploaded file in depth from the following perspectives:
 1. 📋 Overall structure and composition
 2. 📊 Key content and pattern analysis
@@ -103,7 +155,6 @@ def chat(message, history, uploaded_file, model_name, system_message="", max_tok
    - For text/code: Structural features, main patterns
 4. 💡 Potential applications
 5. ✨ Data quality and areas for improvement
-
 Provide detailed and structured analysis from an expert perspective, but explain in an easy-to-understand way. Format the analysis results in Markdown and include specific examples where possible."""
 
     if uploaded_file:
@@ -120,7 +171,6 @@ Provide detailed and structured analysis from an expert perspective, but explain
             
         if message == "Starting file analysis...":
             message = f"""[Structure Analysis] {file_summary}
-
 Please provide detailed analysis from these perspectives:
 1. 📋 Overall file structure and format
 2. 📊 Key content and component analysis
@@ -144,7 +194,7 @@ Please provide detailed analysis from these perspectives:
     messages.append({"role": "user", "content": message})
 
     try:
-        client = get_client(model_name)
+        client = OllamaClient()
         partial_message = ""
         current_history = []
         
@@ -176,13 +226,12 @@ css = """
 footer {visibility: hidden}
 """
 
-
-with gr.Blocks(theme="Yntec/HaleyCH_Theme_Orange", css=css, title="Every RAG 🤖") as demo:
+with gr.Blocks(theme="gstaff/xkcd", css=css, title="Offline Survey Data Analysis 📊") as demo:
     gr.HTML(
         """
         <div style="text-align: center; max-width: 800px; margin: 0 auto;">
-            <h1 style="font-size: 3em; font-weight: 600; margin: 0.5em;">Every RAG Chat 🤖</h1>
-            <h3 style="font-size: 1.2em; margin: 1em;">Your Intelligent File Analysis Assistant 📊</h3>
+            <h1 style="font-size: 3em; font-weight: 600; margin: 0.5em;">Offline Survey Data Analysis</h1>
+            <h3 style="font-size: 1.2em; margin: 1em;">Leveraging Mistral-Nemo via Ollama</h3>
         </div>
         """
     )
@@ -197,7 +246,7 @@ with gr.Blocks(theme="Yntec/HaleyCH_Theme_Orange", css=css, title="Every RAG
             msg = gr.Textbox(
                 label="Type your message",
                 show_label=False,
-                placeholder="Ask me anything about the uploaded file... 💭",
+                placeholder="Ask me anything about the uploaded data file... 💭",
                 container=False
             )
             with gr.Row():
@@ -205,13 +254,6 @@ with gr.Blocks(theme="Yntec/HaleyCH_Theme_Orange", css=css, title="Every RAG
                 send = gr.Button("Send 📤")
         
         with gr.Column(scale=1):
-            model_name = gr.Radio(
-                choices=list(LLM_MODELS.keys()),
-                value="Cohere c4ai-crp-08-2024",
-                label="Select LLM Model 🤖",
-                info="Choose your preferred AI model"
-            )
-            
             gr.Markdown("### Upload File 📁\nSupport: Text, Code, CSV, Parquet files")
             file_upload = gr.File(
                 label="Upload File",
@@ -228,7 +270,7 @@ with gr.Blocks(theme="Yntec/HaleyCH_Theme_Orange", css=css, title="Every RAG
     # Event bindings
     msg.submit(
         chat,
-        inputs=[msg, chatbot, file_upload, model_name, system_message, max_tokens, temperature, top_p],
+        inputs=[msg, chatbot, file_upload, system_message, max_tokens, temperature, top_p],
         outputs=[msg, chatbot],
         queue=True
     ).then(
@@ -239,7 +281,7 @@ with gr.Blocks(theme="Yntec/HaleyCH_Theme_Orange", css=css, title="Every RAG
 
     send.click(
         chat,
-        inputs=[msg, chatbot, file_upload, model_name, system_message, max_tokens, temperature, top_p],
+        inputs=[msg, chatbot, file_upload, system_message, max_tokens, temperature, top_p],
         outputs=[msg, chatbot],
         queue=True
     ).then(
@@ -251,7 +293,7 @@ with gr.Blocks(theme="Yntec/HaleyCH_Theme_Orange", css=css, title="Every RAG
     # Auto-analysis on file upload
     file_upload.change(
         chat,
-        inputs=[gr.Textbox(value="Starting file analysis..."), chatbot, file_upload, model_name, system_message, max_tokens, temperature, top_p],
+        inputs=[gr.Textbox(value="Starting file analysis..."), chatbot, file_upload, system_message, max_tokens, temperature, top_p],
         outputs=[msg, chatbot],
         queue=True
     )