align with paper

app.py

@@ -101,12 +101,9 @@ def generate(
         "max_new_tokens": max_new_tokens,
         "eos_token_id": tokenizer.eos_token_id,
         "early_stopping": True,
+        "no_repeat_ngram_size": 5,
         "repetition_penalty": repetition_penalty,
-        "do_sample": True,
-        "top_p": top_p,
-        "top_k": top_k,
-        "temperature": temperature,
-        "num_beams": 1,
+        "do_sample": False,
     }
 
     t = Thread(target=reft_model.generate, kwargs=generate_kwargs)
@@ -128,33 +125,12 @@ chat_interface = gr.ChatInterface(
             step=1,
             value=DEFAULT_MAX_NEW_TOKENS,
         ),
-        gr.Slider(
-            label="Temperature",
-            minimum=0.1,
-            maximum=2.0,
-            step=0.1,
-            value=0.3,
-        ),
-        gr.Slider(
-            label="Top-p (nucleus sampling)",
-            minimum=0.05,
-            maximum=1.0,
-            step=0.05,
-            value=0.9,
-        ),
-        gr.Slider(
-            label="Top-k",
-            minimum=1,
-            maximum=1000,
-            step=1,
-            value=50,
-        ),
         gr.Slider(
             label="Repetition penalty",
             minimum=1.0,
             maximum=2.0,
             step=0.05,
-            value=1.2,
+            value=1.1,
         ),
     ],
     stop_btn=None,