Update app.py

app.py

@@ -1,25 +1,37 @@
 import torch
-from transformers import T5ForConditionalGeneration,T5Tokenizer, AutoTokenizer, AutoModelForSeq2SeqLM, PegasusTokenizer, PegasusForConditionalGeneration
+from transformers import T5ForConditionalGeneration,T5Tokenizer, AutoTokenizer, AutoModelForSeq2SeqLM
 import streamlit as st
+model_name = st.text_input("Pick a Model", "seduerr/t5-pawraphrase")
+tokenizer = AutoTokenizer.from_pretrained(model_name)  
+model = AutoModelForSeq2SeqLM.from_pretrained("t5-base")
 
+device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
+model = model.to(device)
+def translate_to_english(model, tokenizer, text):
+  translated_text = []
+  text = "paraphrase: " + text + " </s>"
+  encoding = tokenizer.encode_plus(text,pad_to_max_length=True, return_tensors="pt")
+  input_ids, attention_masks = encoding["input_ids"].to(device), encoding["attention_mask"].to(device)
+  beam_outputs = model.generate(
+    input_ids=input_ids, attention_mask=attention_masks,
+    do_sample=True,
+    max_length=256,
+    top_k=120,
+    top_p=0.98,
+    early_stopping=True,
+    num_return_sequences=10
+  )
+  for beam_output in beam_outputs:
+    sent = tokenizer.decode(beam_output, skip_special_tokens=True,clean_up_tokenization_spaces=True)
+    print(sent)
+    translated_text.append(sent)
+  return translated_text
+    
 st.title("Auto Translate (To English)")
 text = st.text_input("Okay")
 st.text("What you wrote: ")
 st.write(text)
 st.text("English Translation: ")
-
-model_name = 'tuner007/pegasus_paraphrase'
-tokenizer = PegasusTokenizer.from_pretrained(model_name)
-model = PegasusForConditionalGeneration.from_pretrained(model_name)
-device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
-model = model.to(device)
-
-def get_response(text):
-  batch = tokenizer([text],truncation=True,padding='longest',max_length=60, return_tensors="pt").to(device)
-  translated = model.generate(**batch,max_length=60, do_sample=True, num_return_sequences=10, temperature=1.5)
-  tgt_text = tokenizer.batch_decode(translated, skip_special_tokens=True)
-  return tgt_text
-    
 if text:
-    translated_text = get_response(text)
-    st.write(translated_text if translated_text else "No translation found")
+    translated_text = translate_to_english(model, tokenizer, text)
+    st.write(translated_text if translated_text else "No translation found")