Update app.py

app.py

@@ -1,11 +1,45 @@
 import gradio as gr
+from transformers import AutoModelForSeq2SeqLM, AutoTokenizer
 
-def respond(system_message, max_length, min_length, max_new_tokens, num_beams, temperature, top_p):
+# Load the model and tokenizer
+model = AutoModelForSeq2SeqLM.from_pretrained("vennify/t5-base-grammar-correction")
+tokenizer = AutoTokenizer.from_pretrained("vennify/t5-base-grammar-correction")
+
+def correct_text(text, max_length, min_length, max_new_tokens, num_beams, temperature, top_p):
+    inputs = tokenizer.encode("grammar: " + text, return_tensors="pt")
+    
+    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(prompt, max_length, min_length, max_new_tokens, num_beams, temperature, top_p):
     # Your response generation logic here
     #response = correct_text(message, max_length, max_new_tokens, min_length, num_beams, temperature, top_p)
     #yield response
     #return f"System message: {system_message}, Max Length: {max_length}, Min Length: {min_length}, Max new tokens: {max_new_tokens}, Num Beams: {num_beams}, Temperature: {temperature}, Top-p: {top_p}"
-    return message
+    #response = correct_text(prompt, max_length, max_new_tokens, min_length, num_beams, temperature, top_p)
+    return prompt
 
 
 # Create the Gradio interface
@@ -17,7 +51,6 @@ with gr.Blocks() as demo:
     prompt_box = gr.Textbox(lines=2, placeholder="Enter your prompt here...")
     output_box = gr.Textbox()
     submitBtn = gr.Button("Submit")
-    
 
     with gr.Accordion("Generation Parameters:", open=False):
         max_length = gr.Slider(minimum=1, maximum=256, value=80, step=1, label="Max Length")
@@ -27,11 +60,11 @@ with gr.Blocks() as demo:
         temperature = gr.Slider(minimum=0.1, maximum=4.0, value=0.7, step=0.1, label="Temperature")
         top_p = gr.Slider(minimum=0.1, maximum=1.0, value=0.95, step=0.05, label="Top-p (nucleus sampling)")
 
-    #show_top_p = gr.Checkbox(value=True, label="Show Top-p Slider")
-    #top_p_slider = gr.Slider(minimum=0.1, maximum=1.0, value=0.95, step=0.05, label="Top-p (nucleus sampling)", visible=True)
-    #show_top_p.change(lambda show: gr.update(visible=show), show_top_p, top_p_slider)
+    show_top_p = gr.Checkbox(value=True, label="Show Top-p Slider")
+    top_p_slider = gr.Slider(minimum=0.1, maximum=1.0, value=0.95, step=0.05, label="Top-p (nucleus sampling)", visible=True)
+    show_top_p.change(lambda show: gr.update(visible=show), show_top_p, top_p_slider)
 
-    submitBtn.click(respond, [prompt_box, max_length, min_length, max_tokens, num_beams, temperature, top_p], output_box)
+    submitBtn.click(correct_text, [prompt_box, max_length, min_length, max_tokens, num_beams, temperature, top_p], output_box)