Update app.py

app.py

@@ -9,6 +9,8 @@ from threading import Thread
 
 print(f"Starting to load the model to memory")
 tok = AutoTokenizer.from_pretrained("stabilityai/stablelm-2-zephyr-1_6b", trust_remote_code=True)
+tok.padding_side = "left"
+tok.pad_token_id = tokenizer.eos_token_id
 # using CUDA for an optimal experience
 slot="<slot_for_user_input_design_by_xm>"
 chat=[{"role": "user", "content": slot}]
@@ -16,7 +18,7 @@ sample_input = tok.apply_chat_template(chat, tokenize=False, add_generation_prom
 input_start_id=sample_input.find(slot)
 prefix=sample_input[:input_start_id]
 suffix=sample_input[input_start_id+len(slot):]
-print(tok.decode(tok.encode(prefix,return_tensors="pt")[0]))
+print(tok.encode(prefix,return_tensors="pt")[0])
 device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')
 m = AutoModelForCausalLM.from_pretrained(
     "stabilityai/stablelm-2-zephyr-1_6b", torch_dtype=torch.float16 if torch.cuda.is_available() else torch.float32, trust_remote_code=True)