Upload app.py

app.py

@@ -23,11 +23,10 @@ model.load_adapter(adapter_path)
 def inference(prompt):
 
   pipe = pipeline(task="text-generation",model=model,tokenizer=tokenizer,max_length = 100)
-  #result = pipe(f"<s>[INST] {prompt} [/INST]")
-  result = pipe(f"{prompt}")
+  result = pipe(f"<s>[INST] {prompt} [/INST]")
   return result[0]['generated_text']
 
-INTERFACE = gr.Interface(fn=inference, inputs=[gr.Textbox(label= "Prompt", value= 'who were The Beatles')],
+INTERFACE = gr.Interface(fn=inference, inputs=[gr.Textbox(label= "Prompt", value= 'what should we do to save time')],
                     
                     outputs=gr.Text(label= "Generated Text"), title="Language Model Phi-2 fine-tuned with OpenAssistant/oasst-1 dataset using QLoRA strategy",