Update app.py

app.py

@@ -1,35 +1,30 @@
 import gradio as gr
-from unsloth import FastLanguageModel
+from transformers import AutoTokenizer, AutoModelForCausalLM, BitsAndBytesConfig
 import torch
 import time
 
 # ----------------------------
 # 🚀 Load Model (CPU)
 # ----------------------------
-def load_model():
-    max_seq_length = 1024
-    model, tokenizer = FastLanguageModel.from_pretrained(
-        model_name="umarfarzan/my-finetuned-model2-lora",
-        max_seq_length=max_seq_length,
-        dtype=None,
-        load_in_4bit=True  # 4-bit works on CPU too
-    )
-    FastLanguageModel.for_inference(model)
-    return model, tokenizer
+model_name = "umarfarzan/my-finetuned-model2-lora"
+
+# LoRA + 4-bit config
+bnb_config = BitsAndBytesConfig(load_in_4bit=True)
 
 print("Loading model...")
-model, tokenizer = load_model()
+tokenizer = AutoTokenizer.from_pretrained(model_name)
+model = AutoModelForCausalLM.from_pretrained(
+    model_name,
+    device_map="cpu",
+    quantization_config=bnb_config
+)
+model.eval()
 print("✅ Model loaded successfully!")
 
 # ----------------------------
 # 💡 Generate Training Program
 # ----------------------------
-def generate_training_program(
-    instruction,
-    max_tokens=500,
-    temperature=0.7,
-    top_p=0.9
-):
+def generate_training_program(instruction, max_tokens=500, temperature=0.7, top_p=0.9):
     prompt_text = f"""Below is an instruction that describes a task, paired with an input that provides further context. Write a response that appropriately completes the request.
 
 ### Instruction:
@@ -52,7 +47,6 @@ def generate_training_program(
 
     generated_text = tokenizer.batch_decode(outputs, skip_special_tokens=True)[0]
 
-    # Extract only the response part
     if "### Response:" in generated_text:
         response = generated_text.split("### Response:")[-1].strip()
     else:
@@ -61,58 +55,23 @@ def generate_training_program(
     return response, f"⏱️ Generated in {generation_time:.2f} seconds"
 
 # ----------------------------
-# 🎨 Gradio Interface
+# 🎨 Gradio UI
 # ----------------------------
-custom_css = """
-.gradio-container {
-    font-family: 'Inter', sans-serif;
-}
-.main-header {
-    text-align: center;
-    background: linear-gradient(135deg, #667eea 0%, #764ba2 100%);
-    color: white;
-    padding: 2rem;
-    border-radius: 10px;
-    margin-bottom: 2rem;
-}
-"""
-
 examples = [
-    ["Design a detailed 1-week training program titled 'The Leader's Blueprint for Strategic Problem-Solving' for mid-level to senior-level managers, team leads, and high-potential employees."],
+    ["Design a detailed 1-week training program titled 'The Leader's Blueprint for Strategic Problem-Solving' for mid-level to senior-level managers."],
     ["Create a 3-day workshop on effective communication skills for remote teams."],
     ["Develop a 5-day leadership bootcamp for new managers focusing on team management and conflict resolution."],
-    ["Design a half-day training session on data-driven decision making for executives."],
-    ["Create a 2-week onboarding program for new software engineers including technical and cultural training."]
 ]
 
-with gr.Blocks(css=custom_css, theme=gr.themes.Soft()) as demo:
-    gr.HTML("""
-        <div class="main-header">
-            <h1>🎯 AI Training Program Generator</h1>
-            <p>Generate professional training programs instantly using AI</p>
-        </div>
-    """)
-
-    with gr.Row():
-        with gr.Column(scale=1):
-            instruction_input = gr.Textbox(
-                label="📝 Training Program Description",
-                placeholder="Describe your training program...",
-                lines=5
-            )
-
-            with gr.Accordion("⚙️ Advanced Settings", open=False):
-                max_tokens_slider = gr.Slider(500, 2000, value=500, step=100, label="Max Output Length")
-                temperature_slider = gr.Slider(0.1, 1.5, value=0.7, step=0.1, label="Creativity (Temperature)")
-                top_p_slider = gr.Slider(0.5, 1.0, value=0.9, step=0.05, label="Diversity (Top-p)")
-
-            generate_btn = gr.Button("🚀 Generate Training Program", variant="primary")
-
-        with gr.Column(scale=1):
-            output_text = gr.Textbox(label="📋 Generated Training Program", lines=25, show_copy_button=True)
-            generation_info = gr.Textbox(label="ℹ️ Generation Info", interactive=False, show_label=False)
-
-    gr.Examples(examples=examples, inputs=[instruction_input], label="💡 Example Prompts")
+with gr.Blocks() as demo:
+    gr.Markdown("## 🎯 AI Training Program Generator")
+    instruction_input = gr.Textbox(label="Training Program Description", lines=5)
+    max_tokens_slider = gr.Slider(500, 2000, value=500, step=100, label="Max Output Length")
+    temperature_slider = gr.Slider(0.1, 1.5, value=0.7, step=0.1, label="Creativity (Temperature)")
+    top_p_slider = gr.Slider(0.5, 1.0, value=0.9, step=0.05, label="Diversity (Top-p)")
+    generate_btn = gr.Button("🚀 Generate Training Program")
+    output_text = gr.Textbox(label="Generated Training Program", lines=25, show_copy_button=True)
+    generation_info = gr.Textbox(label="Generation Info", interactive=False, show_label=False)
 
     generate_btn.click(
         fn=generate_training_program,
@@ -120,13 +79,6 @@ with gr.Blocks(css=custom_css, theme=gr.themes.Soft()) as demo:
         outputs=[output_text, generation_info]
     )
 
-# ----------------------------
-# 🚀 Launch App
-# ----------------------------
-if __name__ == "__main__":
-    demo.queue(max_size=10)
-    demo.launch(
-        share=True,
-        server_name="0.0.0.0",
-        server_port=7860
-    )
+    gr.Examples(examples=examples, inputs=[instruction_input])
+
+demo.launch(share=True)