update to new translation system

app.py

@@ -10,15 +10,18 @@ import os
 
 st.write('Part 1')
 
-question = st.text_input('Posez votre question')
+question_fr = st.text_input('Posez votre question')
 url = st.text_input('mettez le liens de votre image')
 
 if st.button('générer'):
-    responseBase = image(url, question)
+
+    question_en = frenchtoenglish(question_fr)
+
+    responseBase = image(url, question_en)
     st.write('response is :', responseBase)
 
     st.write('Part 2')
-    st.write(longText(responseBase, question))
+    longResponseEn = longText(responseBase, question_en)
+    st.write(englishtofrench(longResponseEn))
 
-    print("####TEST TTS####")
-    os.system("text-to-speach.py")
+    print("####TEST TTS####")

translation.py

@@ -1,30 +1,28 @@
-from transformers import AutoTokenizer, AutoModelWithLMHead, TranslationPipeline, pipeline
+from transformers import M2M100ForConditionalGeneration, M2M100Tokenizer
 
-model_checkpoint = "Helsinki-NLP/opus-mt-en-fr"
-translator = pipeline("translation", model=model_checkpoint)
-# print(translator("how old are you"))
+fr_text = "j'aime coder, j'éspère que vous aussi."
+en_text = "i love code, i hope you too."
 
-########################################################################################
+model = M2M100ForConditionalGeneration.from_pretrained("facebook/m2m100_418M")
+tokenizer = M2M100Tokenizer.from_pretrained("facebook/m2m100_418M")
 
-from transformers import AutoTokenizer, AutoModelWithLMHead, TranslationPipeline
+def frenchtoenglish(input):
+    # translate French to English
+    tokenizer.src_lang = "fr"
+    encoded_fr = tokenizer(input, return_tensors="pt")
+    generated_tokens = model.generate(**encoded_fr, forced_bos_token_id=tokenizer.get_lang_id("en"))
+    return tokenizer.batch_decode(generated_tokens, skip_special_tokens=True)
 
-pipeline = TranslationPipeline(
-    model=AutoModelWithLMHead.from_pretrained("SEBIS/legal_t5_small_trans_fr_en"),
-    tokenizer=AutoTokenizer.from_pretrained(pretrained_model_name_or_path =
-        "SEBIS/legal_t5_small_trans_fr_en", do_lower_case=False, skip_special_tokens=True), device=0)
+def englishtofrench(input):
+    # translate English to French
+    tokenizer.src_lang = "en"
+    encoded_en = tokenizer(input, return_tensors="pt")
+    generated_tokens = model.generate(**encoded_en, forced_bos_token_id=tokenizer.get_lang_id("fr"))
+    return tokenizer.batch_decode(generated_tokens, skip_special_tokens=True)
 
-fr_text = "salut, comment vas-tu ?"
 
-translator2 = pipeline([fr_text], max_length=512)
+### VOICI LES 2 FONCTIONS QUE TU AVAIS FAITES AVANT MAIS QUE J'AI DU REFAIRE
 
-print(translator2)
-
-########################################################################################
-responseBase = "this is the second test"
-
-def englishtofrench():
-    print(translator(responseBase))
-    return translator(responseBase)
-
-englishtofrench()
 
+print(frenchtoenglish(fr_text))
+print(englishtofrench(en_text))