Update src/streamlit_app.py

src/streamlit_app.py

@@ -25,7 +25,7 @@ def load_data():
         "src/BITS_INTERNS.xlsx",
         "src/Summer of AI - ICFAI  (Responses) (3).xlsx"
     ]
-    
+
     combined_df = pd.DataFrame()
     for path in paths:
         try:
@@ -36,39 +36,34 @@ def load_data():
             return None, f"Excel file '{path}' not found. Please upload the file."
         except Exception as e:
             return None, f"Error loading '{path}': {str(e)}"
-    
+
     if combined_df.empty:
         return None, "No data found in either file."
-    
-    return combined_df, None
 
+    return combined_df, None
 
 def extract_numeric_ram(ram) -> Optional[int]:
     if pd.isna(ram):
         return None
-    
+
     ram_str = str(ram).lower().replace(" ", "")
-    
-    # Handle GB format
+
     gb_match = re.search(r"(\d+(?:\.\d+)?)(?:gb|g)", ram_str)
     if gb_match:
         return int(float(gb_match.group(1)))
-    
-    # Handle MB format
+
     mb_match = re.search(r"(\d+)(?:mb|m)", ram_str)
     if mb_match:
         return max(1, int(int(mb_match.group(1)) / 1024))
-    
-    # Handle plain numbers
+
     plain_match = re.search(r"(\d+)", ram_str)
     if plain_match:
         return int(plain_match.group(1))
-    
+
     return None
 
-# Simplified LLM database
 LLM_DATABASE = {
-    "ultra_low": {  # ≤2GB
+    "ultra_low": {
         "general": [
             {"name": "TinyLlama-1.1B-Chat", "size": "637MB", "description": "Compact chat model"},
             {"name": "all-MiniLM-L6-v2", "size": "91MB", "description": "Sentence embeddings"}
@@ -77,7 +72,7 @@ LLM_DATABASE = {
             {"name": "CodeT5-small", "size": "242MB", "description": "Code generation"}
         ]
     },
-    "low": {  # 3-4GB
+    "low": {
         "general": [
             {"name": "Phi-1.5", "size": "2.8GB", "description": "Microsoft's efficient model"},
             {"name": "Gemma-2B", "size": "1.4GB", "description": "Google's compact model"}
@@ -86,7 +81,7 @@ LLM_DATABASE = {
             {"name": "CodeGen-2B", "size": "1.8GB", "description": "Salesforce code model"}
         ]
     },
-    "moderate": {  # 5-8GB
+    "moderate": {
         "general": [
             {"name": "Llama-2-7B-Chat", "size": "3.5GB", "description": "Meta's popular chat model"},
             {"name": "Mistral-7B-Instruct-v0.2", "size": "4.1GB", "description": "Latest Mistral instruct"}
@@ -95,7 +90,7 @@ LLM_DATABASE = {
             {"name": "CodeLlama-7B-Instruct", "size": "3.8GB", "description": "Instruction-tuned CodeLlama"}
         ]
     },
-    "good": {  # 9-16GB
+    "good": {
         "general": [
             {"name": "Llama-2-13B-Chat", "size": "7.3GB", "description": "Larger Llama variant"},
             {"name": "OpenChat-3.5", "size": "7.1GB", "description": "High-quality chat model"}
@@ -104,7 +99,7 @@ LLM_DATABASE = {
             {"name": "CodeLlama-13B-Instruct", "size": "7.3GB", "description": "Larger code model"}
         ]
     },
-    "high": {  # 17-32GB
+    "high": {
         "general": [
             {"name": "Mixtral-8x7B-Instruct-v0.1", "size": "26.9GB", "description": "Mixture of experts"},
             {"name": "Yi-34B-Chat", "size": "19.5GB", "description": "01.AI's large model"}
@@ -113,7 +108,7 @@ LLM_DATABASE = {
             {"name": "CodeLlama-34B-Instruct", "size": "19.0GB", "description": "Large code specialist"}
         ]
     },
-    "ultra_high": {  # >32GB
+    "ultra_high": {
         "general": [
             {"name": "Llama-2-70B", "size": "130GB", "description": "Full precision"},
             {"name": "Mixtral-8x22B", "size": "176GB", "description": "Latest mixture model"}
@@ -122,12 +117,9 @@ LLM_DATABASE = {
 }
 
 def recommend_llm(ram_str) -> Tuple[str, str, Dict[str, List[Dict]]]:
-    """Returns (recommendation, performance_tier, detailed_models)"""
     ram = extract_numeric_ram(ram_str)
-    
     if ram is None:
         return "⚪ Check exact specs", "Unknown", {}
-    
     if ram <= 2:
         return "🔸 Ultra-lightweight models", "Ultra Low", LLM_DATABASE["ultra_low"]
     elif ram <= 4:
@@ -142,10 +134,8 @@ def recommend_llm(ram_str) -> Tuple[str, str, Dict[str, List[Dict]]]:
         return "🔵 Top-tier models", "Ultra High", LLM_DATABASE["ultra_high"]
 
 def get_os_info(os_name) -> Tuple[str, str]:
-    """Returns (icon, clean_name)"""
     if pd.isna(os_name):
-        return "💻", "Not specified"
-    
+        return "🖼", "Not specified"
     os = str(os_name).lower()
     if "windows" in os:
         return "🪟", os_name
@@ -158,17 +148,14 @@ def get_os_info(os_name) -> Tuple[str, str]:
     elif "ios" in os:
         return "📱", os_name
     else:
-        return "💻", os_name
+        return "🖼", os_name
 
 def create_performance_chart(df):
-    """Create RAM distribution chart"""
     laptop_rams = df["Laptop RAM"].apply(extract_numeric_ram).dropna()
     mobile_rams = df["Mobile RAM"].apply(extract_numeric_ram).dropna()
-    
     fig = go.Figure()
     fig.add_trace(go.Histogram(x=laptop_rams, name="Laptop RAM", opacity=0.7))
     fig.add_trace(go.Histogram(x=mobile_rams, name="Mobile RAM", opacity=0.7))
-    
     fig.update_layout(
         title="RAM Distribution",
         xaxis_title="RAM (GB)",
@@ -179,121 +166,96 @@ def create_performance_chart(df):
     return fig
 
 def display_models(models_dict: Dict[str, List[Dict]]):
-    """Display models by category"""
     if not models_dict:
         return
-    
     for category, model_list in models_dict.items():
         if model_list:
             st.markdown(f"**{category.title()} Models:**")
-            for model in model_list[:5]:  # Limit to 5 per category
+            for model in model_list[:5]:
                 st.write(f"• {model['name']} ({model['size']}) - {model['description']}")
 
-# Main App
 st.title("🧠 LLM Compatibility Advisor")
 st.markdown("Get personalized AI model recommendations with download sizes!")
 
-# Load data
 df, error = load_data()
-
 if error:
     st.error(error)
     st.stop()
-
 if df is None or df.empty:
     st.error("No data found.")
     st.stop()
 
-# Sidebar
 with st.sidebar:
     st.header("📊 Quick Stats")
     st.metric("Total Students", len(df))
-    
     avg_laptop_ram = df["Laptop RAM"].apply(extract_numeric_ram).mean()
     avg_mobile_ram = df["Mobile RAM"].apply(extract_numeric_ram).mean()
-    
     if not pd.isna(avg_laptop_ram):
         st.metric("Avg Laptop RAM", f"{avg_laptop_ram:.1f} GB")
     if not pd.isna(avg_mobile_ram):
         st.metric("Avg Mobile RAM", f"{avg_mobile_ram:.1f} GB")
 
-# Individual Analysis
 st.subheader("👤 Individual Student Analysis")
-
-# Fixed selectbox - use index-based selection
 student_names = list(df["Full Name"].unique())
 student_options = ["Select a student..."] + student_names
 
-selected_index = st.selectbox(
+selected_name = st.selectbox(
     "Choose a student:",
-    options=range(len(student_options)),
-    format_func=lambda x: student_options[x]
+    options=student_options,
 )
 
-if selected_index > 0:  # If not the placeholder option
-    selected_user = student_names[selected_index - 1]
+if selected_name != "Select a student...":
+    selected_user = selected_name
     user_data = df[df["Full Name"] == selected_user].iloc[0]
-    
+
     col1, col2 = st.columns(2)
-    
+
     with col1:
         st.markdown("### 💻 Laptop")
         laptop_os_icon, laptop_os_name = get_os_info(user_data.get('Laptop Operating System'))
         laptop_ram = user_data.get('Laptop RAM', 'Not specified')
-        laptop_rec, laptop_tier, laptop_models = recommend_llm(laptop_ram)
-        
+        laptop_rec, _, laptop_models = recommend_llm(laptop_ram)
         st.markdown(f"**OS:** {laptop_os_icon} {laptop_os_name}")
         st.markdown(f"**RAM:** {laptop_ram}")
         st.success(f"**Recommendation:** {laptop_rec}")
-        
         display_models(laptop_models)
-    
+
     with col2:
         st.markdown("### 📱 Mobile")
         mobile_os_icon, mobile_os_name = get_os_info(user_data.get('Mobile Operating System'))
         mobile_ram = user_data.get('Mobile RAM', 'Not specified')
-        mobile_rec, mobile_tier, mobile_models = recommend_llm(mobile_ram)
-        
+        mobile_rec, _, mobile_models = recommend_llm(mobile_ram)
         st.markdown(f"**OS:** {mobile_os_icon} {mobile_os_name}")
         st.markdown(f"**RAM:** {mobile_ram}")
         st.success(f"**Recommendation:** {mobile_rec}")
-        
         display_models(mobile_models)
 
-# Batch Analysis
 st.markdown("---")
 st.header("📊 Batch Analysis")
-
-# Create summary table
 df_display = df[["Full Name", "Laptop RAM", "Mobile RAM"]].copy()
 df_display["Laptop Recommendation"] = df["Laptop RAM"].apply(lambda x: recommend_llm(x)[0])
 df_display["Mobile Recommendation"] = df["Mobile RAM"].apply(lambda x: recommend_llm(x)[0])
-
 st.dataframe(df_display, use_container_width=True)
 
-# Performance chart
 if len(df) > 1:
     st.subheader("📈 RAM Distribution")
     fig = create_performance_chart(df)
     st.plotly_chart(fig, use_container_width=True)
 
-# Model Explorer
 st.markdown("---")
 st.header("🔍 Model Explorer")
-
 selected_ram_range = st.selectbox(
     "Select RAM range:",
-    ["≤2GB (Ultra Low)", "3-4GB (Low)", "5-8GB (Moderate)", 
+    ["\u22642GB (Ultra Low)", "3-4GB (Low)", "5-8GB (Moderate)", 
      "9-16GB (Good)", "17-32GB (High)", ">32GB (Ultra High)"]
 )
 
-# Map selection to database
 ram_mapping = {
     "≤2GB (Ultra Low)": "ultra_low",
     "3-4GB (Low)": "low", 
     "5-8GB (Moderate)": "moderate",
     "9-16GB (Good)": "good",
-    "17-32GB (High)": " high ",
+    "17-32GB (High)": "high",
     ">32GB (Ultra High)": "ultra_high"
 }
 
@@ -302,25 +264,24 @@ if selected_key in LLM_DATABASE:
     st.subheader(f"Models for {selected_ram_range}")
     display_models(LLM_DATABASE[selected_key])
 
-# Quick reference
 with st.expander("📘 Quick Reference"):
     st.markdown("""
     ## Popular Models by Category
-    
+
     **General Purpose:**
     - Llama-2 Series (7B, 13B, 70B)
     - Mistral Series
     - Gemma (2B, 7B)
-    
+
     **Code Specialists:**
     - CodeLlama
     - CodeGen
-    
+
     **Where to Download:**
     - 🤗 Hugging Face Hub
-    - 🦙 Ollama
+    - 🧙️ Ollama
     - 📦 LM Studio
     """)
 
 st.markdown("---")
-st.markdown("*Built for BITS Pilani Interns*")
+st.markdown("*Built for BITS Pilani Interns*")