Update app.py

app.py

@@ -32,9 +32,9 @@ def run_generation(
     temperature,
     top_k,
     max_new_tokens,
-    no_repeat_ngram_size=4,
-    length_penalty=1.0,
     repetition_penalty=1.1,
+    length_penalty=1.0,
+    no_repeat_ngram_size=4,
     use_generation_config=False,
 ):
     st = time.perf_counter()
@@ -55,6 +55,8 @@ def run_generation(
         top_p=top_p,
         temperature=float(temperature),
         top_k=top_k,
+        repetition_penalty=repetition_penalty,
+        length_penalty=length_penalty,
         no_repeat_ngram_size=no_repeat_ngram_size,
     )
     t = Thread(target=model.generate, kwargs=generate_kwargs)
@@ -90,7 +92,7 @@ with gr.Blocks() as demo:
     with gr.Row():
         with gr.Column(scale=4):
             user_text = gr.Textbox(
-                placeholder="Write an email about an alpaca that likes flan",
+                value="Can you write an email to Jerome letting him know that thy're taking the hobbits to Isengard?",
                 label="User input",
             )
             model_output = gr.Textbox(label="Model output", lines=10, interactive=False)
@@ -121,23 +123,38 @@ with gr.Blocks() as demo:
                 interactive=True,
                 label="Top-k",
             )
-            temperature = gr.Slider(
-                minimum=0.1,
-                maximum=5.0,
-                value=0.8,
+            repetition_penalty = gr.Slider(
+                minimum=0.9,
+                maximum=2.5,
+                value=1.1,
                 step=0.1,
                 interactive=True,
-                label="Temperature",
+                label="Repetition Penalty",
             )
-
+            length_penalty = gr.Slider(
+                minimum=0.8,
+                maximum=1.5,
+                value=1.0,
+                step=0.1,
+                interactive=True,
+                label="Length Penalty",
+            )
+            # temperature = gr.Slider(
+            #     minimum=0.1,
+            #     maximum=5.0,
+            #     value=0.8,
+            #     step=0.1,
+            #     interactive=True,
+            #     label="Temperature",
+            # )
     user_text.submit(
         run_generation,
-        [user_text, top_p, temperature, top_k, max_new_tokens],
+        [user_text, top_p, temperature, top_k, max_new_tokens, repetition_penalty, length_penalty],
         model_output,
     )
     button_submit.click(
         run_generation,
-        [user_text, top_p, temperature, top_k, max_new_tokens],
+        [user_text, top_p, temperature, top_k, max_new_tokens, repetition_penalty, length_penalty],
         model_output,
     )