Update app.py

app.py

@@ -190,7 +190,7 @@ def fast_decode_step(token, e_seq, h, c, decoder_model, e_mask=None):
     c = tf.convert_to_tensor(c, dtype=tf.float32)
     return decoder_model([token, e_seq, e_mask, h, c], training=False)
 
-def reply(text, max_decode_len=MAX_LEN, rep_penalty=1.3, beam_width=7, length_penalty=0.7, temperature=0.7):
+def reply(text, max_decode_len=MAX_LEN, rep_penalty=1.3, beam_width=3, length_penalty=0.7, temperature=0.7):
     text = text.strip()
     if MoE:
         if not text.startswith("<TASK_"):
@@ -301,12 +301,12 @@ def respond(message, history, max_tokens, temperature, top_p):
 with gr.Blocks() as demo:
     with gr.Sidebar():
         gr.LoginButton()
-        gr.Markdown("### Model Info\nIAMAM v1.0.0\n40M Parameters\nLSTM + Attention")
+        gr.Markdown("### Model Info\nIAMAM v1.1.0\n40M Parameters\nLSTM + Attention")
 
     gr.ChatInterface(
         fn=respond,
-        title="IAMAM (I AM A Model)",
-        description="A lightweight Keras 3 LSTM Model with bidirectionality and attention! (Powered by Beam Search)",
+        title="A Math AutoBiography maker!",
+        description="Try asking it to make a autobiography on 2 + 2! (Runs IAMAM v1.1.0)",
         additional_inputs=[
             gr.Slider(minimum=1, maximum=MAX_LEN, value=32, step=1, label="Max new tokens"),
             gr.Slider(minimum=0.1, maximum=4.0, value=0.7, step=0.1, label="Temperature"),