🐛 Fix nbr parameter type issue when submitting new models

backend/api.py

@@ -20,7 +20,7 @@ async def submit_model(
     hf_space_tag: str,
     model_description: str,
     organization: str,
-    model_size: str,
+    model_size: int,  # Changed from str to int
     publication_title: str,
     publication_link: str,
 
@@ -31,12 +31,12 @@ async def submit_model(
     n_shot: str,
 ) -> Dict[str, Any]:
     """API endpoint for model submission with rate limiting."""
-    
+
     # Check rate limit
     if not check_submission_limit(hf_space_tag):
         status = get_submission_status(hf_space_tag)
         raise Exception(f"Daily submission limit exceeded. Used {status['submissions_today']}/{status['daily_limit']} submissions today. Try again tomorrow.")
-    
+
     # Get test smiles and labels
     test_smiles, test_labels = load_test_dataset()
 
@@ -46,7 +46,7 @@ async def submit_model(
         hf_space_tag=hf_space_tag,
         model_description=model_description,
         organization=organization or "",
-        model_size=model_size or "",
+        model_size=model_size,  # Now numeric
         publication_title=publication_title or "",
         publication_link=publication_link or "",
 

backend/schema.py

@@ -7,7 +7,7 @@ def create_submission_record(
     hf_space_tag: str,
     model_description: str,
     organization: str,
-    model_size: str,
+    model_size: int,  # Changed from str to int
     publication_title: str,
     publication_link: str,
 
@@ -52,15 +52,15 @@ def create_submission_record(
 
 def get_dataset_schema() -> Dict[str, Any]:
     """Return the HuggingFace dataset schema."""
-    
+
     return {
         "config": {
             "model_name": "string",
-            "hf_space_tag": "string", 
+            "hf_space_tag": "string",
             "model_description": "string",
             "publication_title": "string",
             "publication_link": "string",
-            "model_size": "string",
+            "model_size": "int64",  # Changed from "string" to "int64" for numeric values
             "pretraining": "string",
             "organization": "string",
             "date_submitted": "string",

backend/submission.py

@@ -12,7 +12,7 @@ async def process_submission(
     hf_space_tag: str,
     model_description: str,
     organization: str,
-    model_size: str,
+    model_size: int,  # Changed from str to int
     publication_title: str,
     publication_link: str,
     pretrained: bool,

frontend/leaderboard.py

@@ -10,6 +10,46 @@ from datetime import datetime
 from config.leaderboard import MAX_DECIMALS, COLUMN_NAMES
 
 
+def parse_parameter_count(value):
+    """Parse parameter count from various formats to raw numeric value.
+
+    Accepts:
+        - Raw numbers: 120000000, "120000000"
+        - Human-readable: "120M", "0.12B", "154K"
+        - Empty/None values
+
+    Args:
+        value: Parameter count in any supported format
+
+    Returns:
+        int: Raw parameter count, or None for empty/invalid values
+    """
+    if pd.isna(value) or value == "" or value is None:
+        return None
+
+    # If already a number, return it
+    if isinstance(value, (int, float)):
+        return int(value)
+
+    # Convert string to number
+    value_str = str(value).strip().upper()
+    if not value_str:
+        return None
+
+    # Extract numeric part and suffix
+    import re
+    match = re.match(r'^([0-9.]+)\s*([KMBT]?)$', value_str)
+    if not match:
+        return None
+
+    num_part = float(match.group(1))
+    suffix = match.group(2)
+
+    # Apply multiplier based on suffix
+    multipliers = {'K': 1e3, 'M': 1e6, 'B': 1e9, 'T': 1e12, '': 1}
+    return int(num_part * multipliers[suffix])
+
+
 def format_parameter_count(value):
     """Format parameter count to human-readable string (B, M, K).
 

frontend/submission.py

@@ -1,6 +1,7 @@
 import asyncio
 import gradio as gr
 from backend.api import submit_model
+from frontend.leaderboard import parse_parameter_count
 
 
 def handle_submission(
@@ -22,10 +23,15 @@ def handle_submission(
     # Basic validation
     if not model_name or not hf_space_tag or not model_description or not organization or not model_size or not pretrained or not publication_title or not publication_link or not zero_shot or not few_shot:
         return "<div style='color: red;'>Please fill in all required fields (*)</div>"
-    
+
     if "/" not in hf_space_tag:
         return "<div style='color: red;'>HuggingFace space tag should be in format 'username/space-name'</div>"
-    
+
+    # Parse and validate model_size
+    parsed_model_size = parse_parameter_count(model_size)
+    if model_size and parsed_model_size is None:
+        return "<div style='color: red;'>Invalid model size format. Use raw numbers (e.g., 120000000) or human-readable format (e.g., 120M, 0.12B)</div>"
+
     # Process submission
     try:
         result = asyncio.run(submit_model(
@@ -33,7 +39,7 @@ def handle_submission(
             hf_space_tag=hf_space_tag,
             model_description=model_description,
             organization=organization or "",
-            model_size=model_size or "",
+            model_size=parsed_model_size,
             pretrained = pretrained,
             pretraining_data = pretraining_data or "",
             publication_title=publication_title or "",