Create app.py

app.py

@@ -0,0 +1,19 @@
+import gradio as gr
+from transformers import AutoModelForCausalLM, AutoTokenizer
+from peft import PeftModel
+
+base_model = "microsoft/phi-2"
+adapter_path = "your-username/brahma-lora"  # Replace with your adapter repo
+
+# Load model and LoRA adapter
+tokenizer = AutoTokenizer.from_pretrained(base_model)
+model = AutoModelForCausalLM.from_pretrained(base_model, device_map="auto")
+model = PeftModel.from_pretrained(model, adapter_path)
+
+def generate_response(prompt):
+    inputs = tokenizer(prompt, return_tensors="pt").to(model.device)
+    output = model.generate(**inputs, max_new_tokens=100)
+    return tokenizer.decode(output[0], skip_special_tokens=True)
+
+demo = gr.Interface(fn=generate_response, inputs="text", outputs="text")
+demo.launch()