Update app.py

app.py

@@ -47,9 +47,9 @@ df = pd.read_sql_query(query, conn)
 conn.close()
 '''
 
-'''
+
 # Create a sample DataFrame with 3,000 records and 20 columns
-num_records = 100
+num_records = 3000
 num_columns = 20
 
 data = {
@@ -64,7 +64,7 @@ data["year"] = [random.choice(years) for _ in range(num_records)]
 data["city"] = [random.choice(cities) for _ in range(num_records)]
 
 table = pd.DataFrame(data)
-'''
+
 #table = pd.read_csv(csv_file.name, delimiter=",")
 #table.fillna(0, inplace=True)
 #table = table.astype(str)
@@ -73,7 +73,7 @@ data = {
     "year": [1896, 1900, 1904, 2004, 2008, 2012],
     "city": ["athens", "paris", "st. louis", "athens", "beijing", "london"]
 }
-table = pd.DataFrame.from_dict(data)
+#table = pd.DataFrame.from_dict(data)
 
 
 # Load the chatbot model
@@ -130,7 +130,48 @@ def chat(input, history=[]):
 def sqlquery(input): #, history=[]):
 
     global conversation_history
-    
+
+    #======================================================================
+    batch_size = 10  # Number of records in each batch
+    num_records = 3000  # Total number of records in the dataset
+    for start_idx in range(0, num_records, batch_size):
+        end_idx = min(start_idx + batch_size, num_records)
+        
+        # Get a batch of records
+        batch_data = table[start_idx:end_idx]
+
+        batch_responses = []
+        
+        for idx, record in enumerate(batch_data):
+            # Maintain conversation context by appending history
+            if conversation_history:
+                history = "\n".join(conversation_history)
+                input_text = history + "\nUser: " + record["question"]
+            else:
+                input_text = "User: " + record["question"]
+            
+            # Tokenize the input text
+            tokenized_input = sql_tokenizer.encode(input_text, return_tensors="pt")
+            
+            # Perform inference
+            with torch.no_grad():
+                output = sql_model.generate(
+                    input_ids=tokenized_input,
+                    max_length=1024,
+                    pad_token_id=sql_tokenizer.eos_token_id,
+                )
+            
+            # Decode the output and process the response
+            response = sql_tokenizer.decode(output[0], skip_special_tokens=True)
+            batch_responses.append(response)
+
+             # Update conversation history
+            conversation_history.append("User: " + record["question"])
+            conversation_history.append("Bot: " + response)
+
+            
+    # ==========================================================================
+    '''
     inputs = [input]
     sql_encoding = sql_tokenizer(table=table, query=input, return_tensors="pt")
     sql_outputs = sql_model.generate(**sql_encoding)
@@ -139,7 +180,7 @@ def sqlquery(input): #, history=[]):
     #history.append((input, sql_response))
     conversation_history.append(("User", input))
     conversation_history.append(("Bot", sql_response))
-
+    '''
     # Build conversation string
     #conversation = "\n".join([f"User: {user_msg}\nBot: {resp_msg}" for user_msg, resp_msg in conversation_history])
     conversation = "\n".join([f"{sender}: {msg}" for sender, msg in conversation_history])