Upload 2 files

app.py

@@ -1,4 +1,3 @@
-import spaces
 import gradio as gr
 from transformers import AutoTokenizer, AutoModelForSeq2SeqLM
 import spacy
@@ -18,22 +17,20 @@ class ModelSingleton:
 
             # Load translation models and tokenizers
             self.tokenizer_en_it = AutoTokenizer.from_pretrained("LeonardPuettmann/Quadrifoglio-mt-en-it")
-            self.model_en_it = AutoModelForSeq2SeqLM.from_pretrained("LeonardPuettmann/Quadrifoglio-mt-en-it")
+            self.model_en_it = AutoModelForSeq2SeqLM.from_pretrained("LeonardPuettmann/Quadrifoglio-mt-en-it", load_in_8bit=True)
 
             self.tokenizer_it_en = AutoTokenizer.from_pretrained("LeonardPuettmann/Quadrifoglio-mt-it-en")
-            self.model_it_en = AutoModelForSeq2SeqLM.from_pretrained("LeonardPuettmann/Quadrifoglio-mt-it-en")
+            self.model_it_en = AutoModelForSeq2SeqLM.from_pretrained("LeonardPuettmann/Quadrifoglio-mt-it-en", load_in_8bit=True)
 
             self.initialized = True
 
 model_singleton = ModelSingleton()
 
-@spaces.GPU(duration=120)
 def generate_response_en_it(input_text):
     input_ids = model_singleton.tokenizer_en_it("translate English to Italian: " + input_text, return_tensors="pt").input_ids
     output = model_singleton.model_en_it.generate(input_ids, max_new_tokens=256)
     return model_singleton.tokenizer_en_it.decode(output[0], skip_special_tokens=True)
 
-@spaces.GPU(duration=120)
 def generate_response_it_en(input_text):
     input_ids = model_singleton.tokenizer_it_en("translate Italian to English: " + input_text, return_tensors="pt").input_ids
     output = model_singleton.model_it_en.generate(input_ids, max_new_tokens=256)

requirements.txt

@@ -1,7 +1,6 @@
 gradio
 transformers
 spacy
-spaces
 torch
 en-core-web-sm @ https://github.com/explosion/spacy-models/releases/download/en_core_web_sm-3.7.1/en_core_web_sm-3.7.1-py3-none-any.whl
 it-core-news-sm @ https://github.com/explosion/spacy-models/releases/download/it_core_news_sm-3.8.0/it_core_news_sm-3.8.0-py3-none-any.whl