Update src/streamlit_app.py

src/streamlit_app.py

@@ -51,49 +51,6 @@ def load_data():
 def extract_numeric_ram(ram) -> Optional[int]:
     if pd.isna(ram):
         return None
-def calculate_quantized_size(base_size_str: str, quantization: str) -> str:
-    """Calculate quantized model size based on base size and quantization type"""
-    try:
-        # Extract numeric value and unit from base size
-        import re
-        match = re.match(r'(\d+(?:\.\d+)?)\s*(GB|MB)', base_size_str.upper())
-        if not match:
-            return base_size_str
-        
-        value, unit = float(match.group(1)), match.group(2)
-        multiplier = QUANTIZATION_INFO[quantization]["multiplier"]
-        
-        # Calculate new size
-        new_value = value * multiplier
-        
-        # Convert MB to GB if needed for better readability
-        if unit == "MB" and new_value >= 1024:
-            new_value = new_value / 1024
-            unit = "GB"
-        elif unit == "GB" and new_value < 1:
-            new_value = new_value * 1024
-            unit = "MB"
-        
-        # Format the result
-        if new_value >= 10:
-            return f"{new_value:.0f}{unit}"
-        else:
-            return f"{new_value:.1f}{unit}"
-    except:
-        return base_size_str
-
-def get_quantization_recommendations(ram_gb: int) -> List[str]:
-    """Recommend best quantization options based on available RAM"""
-    if ram_gb <= 2:
-        return ["4bit"]
-    elif ram_gb <= 4:
-        return ["4bit", "8bit"]
-    elif ram_gb <= 8:
-        return ["4bit", "8bit"]
-    elif ram_gb <= 16:
-        return ["8bit", "fp16"]
-    else:
-        return ["fp16", "8bit", "4bit"]
     
     ram_str = str(ram).lower().replace(" ", "")
     
@@ -115,14 +72,7 @@ def get_quantization_recommendations(ram_gb: int) -> List[str]:
     return None
 
 # Streamlined LLM database with popular models and download sizes
-
-QUANTIZATION_INFO = {
-    "fp16": {"multiplier": 1.0, "label": "FP16 (Full)", "description": "Best quality, largest size"},
-    "8bit": {"multiplier": 0.5, "label": "8-bit", "description": "Good quality, 50% smaller"},
-    "4bit": {"multiplier": 0.25, "label": "4-bit", "description": "Decent quality, 75% smaller"}
-}
 LLM_DATABASE = {
-    
     "ultra_low": {  # ≤2GB
         "general": [
             {"name": "TinyLlama-1.1B-Chat", "size": "637MB", "description": "Compact chat model"},
@@ -233,7 +183,6 @@ LLM_DATABASE = {
     }
 }
 
-
 # Enhanced LLM recommendation with performance tiers
 def recommend_llm(ram_str) -> Tuple[str, str, str, Dict[str, List[Dict]]]:
     """Returns (recommendation, performance_tier, additional_info, detailed_models)"""