Update app.py

app.py

@@ -10,21 +10,34 @@ model_name = "vennify/t5-base-grammar-correction"
 model = AutoModelForSeq2SeqLM.from_pretrained(model_name)
 tokenizer = AutoTokenizer.from_pretrained(model_name)
 
-def correct_text(text, max_length, num_beams, temperature, top_p):
+def correct_text(text, max_length, max_new_tokens=0, min_length, num_beams, temperature, top_p):
     inputs = tokenizer.encode(text, return_tensors="pt")
-    outputs = model.generate(
-        inputs,
-        max_length=max_length,
-        num_beams=num_beams,
-        temperature=temperature,
-        top_p=top_p,
-        early_stopping=True
-    )
+    if max_new_tokens > 0:
+        outputs = model.generate(
+            inputs,
+            max_length=max_length,
+            max_new_tokens=max_new_tokens,
+            min_length=min_length,
+            num_beams=num_beams,
+            temperature=temperature,
+            top_p=top_p,
+            early_stopping=True
+        )
+    else:
+        outputs = model.generate(
+            inputs,
+            max_length=max_length,
+            min_length=min_length,
+            num_beams=num_beams,
+            temperature=temperature,
+            top_p=top_p,
+            early_stopping=True
+        )
     corrected_text = tokenizer.decode(outputs[0], skip_special_tokens=True)
     return corrected_text
 
 
-def respond(message, history: list[tuple[str, str]], system_message, max_length, min_length, num_beams, temperature, top_p):
+def respond(message, history: list[tuple[str, str]], system_message, max_length, min_length, max_new_tokens, num_beams, temperature, top_p):
     #messages = [{"role": "system", "content": system_message}]
 
     #for val in history:
@@ -35,7 +48,7 @@ def respond(message, history: list[tuple[str, str]], system_message, max_length,
 
     #messages.append({"role": "user", "content": message})
 
-    response = correct_text(message, max_length, min_length, num_beams, temperature, top_p)
+    response = correct_text(message, max_length, max_new_tokens, min_length, num_beams, temperature, top_p)
     yield response
 
 """
@@ -47,6 +60,7 @@ demo = gr.ChatInterface(
         gr.Textbox(value="You are a friendly Chatbot.", label="System message"),
         gr.Slider(minimum=1,   maximum=2048, value=100,  step=1,    label="Max Length"),
         gr.Slider(minimum=1,   maximum=2048, value=0,    step=1,    label="Min Length"),
+        gr.Slider(minimum=1,   maximum=2048, value=0,    step=1,    label="Max New Tokens (optional)"),
         gr.Slider(minimum=1,   maximum=10,   value=5,    step=1,    label="Num Beams"),
         gr.Slider(minimum=0.1, maximum=4.0,  value=0.7,  step=0.1,  label="Temperature"),
         gr.Slider(minimum=0.1, maximum=1.0,  value=0.95, step=0.05, label="Top-p (nucleus sampling)"),