removed spider folder and updated code

app.py

@@ -3,7 +3,6 @@ import random
 import time
 import json
 import gradio as gr
-from datasets import Dataset
 from huggingface_hub import InferenceClient
 
 # ------------------------
@@ -12,17 +11,10 @@ from huggingface_hub import InferenceClient
 HF_TOKEN = os.environ.get("HF_TOKEN")
 
 # ------------------------
-# Clone Spider Dataset if Not Exists
+# Load Spider Dataset (local JSON)
 # ------------------------
-if not os.path.exists("spider/train_spider.json"):
-    os.system("git clone https://github.com/taoyds/spider.git")
-
-# ------------------------
-# Load Spider Dataset
-# ------------------------
-with open("spider/train_spider.json", "r") as f:
-    spider_raw = json.load(f)
-spider_dataset = Dataset.from_list(spider_raw)
+with open("train_spider.json", "r") as f:
+    spider_dataset = json.load(f)
 
 # ------------------------
 # Inference Clients
@@ -42,7 +34,7 @@ model_list = {
 }
 
 # ------------------------
-# Prompt Engineering Template
+# Prompt Template for SQL Generation
 # ------------------------
 def build_prompt(user_question):
     return f"""You are an expert SQL assistant. Convert the given question into a valid SQL query.
@@ -50,7 +42,7 @@ def build_prompt(user_question):
 Instructions:
 - Respond with only the SQL query.
 - Do not include markdown, explanations, or additional formatting.
-- Use correct table and column names from the schema.
+- Use correct table and column names.
 - Follow SQL best practices and Spider dataset formatting.
 
 Examples:
@@ -70,15 +62,16 @@ A:"""
 def evaluate_all_models(user_input, expected_sql, chat_history):
     evaluations = []
     full_chat_transcript = ""
-    engineered_prompt = build_prompt(user_input)
+    prompt = build_prompt(user_input)
 
     for model_name, model_config in model_list.items():
         client = model_config["client"]
         model_id = model_config["model_id"]
 
-        messages = chat_history + [{"role": "user", "content": engineered_prompt}]
+        messages = chat_history + [{"role": "user", "content": prompt}]
         try:
             start_time = time.time()
+
             result = client.chat.completions.create(
                 model=model_id,
                 messages=messages
@@ -106,6 +99,7 @@ def evaluate_all_models(user_input, expected_sql, chat_history):
             f"- Response Latency: {latency} ms ({latency_status})\n"
         )
         evaluations.append(summary)
+
         full_chat_transcript += f"\n👤 User: {user_input}\n🤖 {model_name}: {model_sql}\n"
 
     return full_chat_transcript.strip(), chat_history, "\n\n".join(evaluations)
@@ -121,7 +115,7 @@ def get_random_spider_prompt():
 # Gradio UI
 # ------------------------
 with gr.Blocks() as demo:
-    gr.Markdown("## 🧠 Spider Dataset Model Evaluation with Prompt Engineering")
+    gr.Markdown("## 🧠 Spider Dataset Model Evaluation")
 
     prompt_input = gr.Textbox(label="Your Prompt", lines=3, placeholder="Ask your BI question...")
     expected_sql_display = gr.Textbox(label="Expected SQL", lines=2, interactive=False)
@@ -147,5 +141,5 @@ with gr.Blocks() as demo:
         outputs=[chat_display, chat_memory, evaluation_display]
     )
 
-# Launch the app
+# Launch
 demo.launch()

spider

@@ -1 +0,0 @@
-Subproject commit b7b5b8c890cd30e35427348bb9eb8c6d1350ca7c

train_spider.json

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:c43d0d72e59e1a9e1a60837da9bf70d5a6277226bdb7f634d544f380646f527a
-size 24928884