Update app.py

app.py

@@ -18,11 +18,21 @@ def run_inference_on_model(prompt):
     return (result[0]['generated_text'].replace(prompt, ''))
     
 
-model_path = './fine_tuned_model'
-model = AutoModelForCausalLM.from_pretrained(model_path, trust_remote_code=True, local_files_only=True)
-tokenizer = AutoTokenizer.from_pretrained(model_path)
+
 max_length = 256
 
+# Load Model
+model_name = "microsoft/phi-2"
+model = AutoModelForCausalLM.from_pretrained(model_name,trust_remote_code=True)
+model.config.use_cache = False
+# Load tokenizer
+tokenizer = AutoTokenizer.from_pretrained(model_name, trust_remote_code=True)
+tokenizer.pad_token = tokenizer.eos_token
+# Load the fine tuned weights
+finetuned_model_path = './fine_tuned_model'
+model.load_adapter(finetuned_model_path)
+
+
 # Define Interface
 description = 'An AI assistant that runs on the Microsoft Phi 2 model fine tuned on Open Assistant dataset using QLora approach. Link to the model: https://huggingface.co/microsoft/phi-2 Link to the dataset: https://huggingface.co/datasets/OpenAssistant/oasst1  '
 title = 'AI Chat bot finetuned on Microsoft Phi 2 model'