Update app.py

app.py

@@ -19,6 +19,7 @@ text = '<|im_start|><|im_end|>user\n'+inner+'<|im_end|>\n<|im_start|>assistant\n
 inputs = tokenizer(text, return_tensors="pt")
 outputs = model.generate(**inputs, max_new_tokens=token_step_size)
 
-while (not torch.any(outputs[0][-token_step_size:] == 4)):
-    outputs = model.generate(input_ids=outputs, attention_mask=torch.ones_like(outputs),max_new_tokens=token_step_size)
-    st.write(tokenizer.decode(outputs[0][-token_step_size:], skip_special_tokens=True))#, end=' ', flush=True)
+if inner:
+    while (not torch.any(outputs[0][-token_step_size:] == 4)):
+        outputs = model.generate(input_ids=outputs, attention_mask=torch.ones_like(outputs),max_new_tokens=token_step_size)
+        st.write(tokenizer.decode(outputs[0][-token_step_size:], skip_special_tokens=True))#, end=' ', flush=True)