~ app.py

app.py

@@ -1,27 +1,50 @@
-import gradio as gr
+# import gradio as gr
+
+# from transformers import AutoModelForSeq2SeqLM, AutoTokenizer
+
+# model_path = "anzorq/m2m100_418M_ft_ru-kbd_44K"  
+# src_lang="ru" 
+# tgt_lang="zu"
+
+# tokenizer = AutoTokenizer.from_pretrained(model_path, src_lang=src_lang)
+# model = AutoModelForSeq2SeqLM.from_pretrained(model_path)
+
+# def translate(text):
+#   inputs = tokenizer(text, return_tensors="pt")
 
-from transformers import AutoModelForSeq2SeqLM, AutoTokenizer
+#   translated_tokens = model.generate(
+#       **inputs, forced_bos_token_id=tokenizer.lang_code_to_id[tgt_lang], num_beams=4, num_return_sequences=4
+#   )
+
+#   translations = []
+#   for translation in tokenizer.batch_decode(translated_tokens, skip_special_tokens=True):
+#       translations.append(translation)
+
+#   return translations
+
+# output = gr.outputs.Textbox()
+# iface = gr.Interface(fn=translate, inputs="text", outputs=output)
+# iface.launch()
+
+import gradio as gr
 
-model_path = "anzorq/m2m100_418M_ft_ru-kbd_44K"  
-src_lang="ru" 
-tgt_lang="zu"
+title = "Русско-черкесский переводчик"
+description = """
+              Demo of a Russian-Circassian (Kabardian dialect) translator. 
 
-tokenizer = AutoTokenizer.from_pretrained(model_path, src_lang=src_lang)
-model = AutoModelForSeq2SeqLM.from_pretrained(model_path)
+              The translator is based on a machine learning model that has been trained on 45,000 Russian-Circassian sentence pairs. 
 
-def translate(text):
-  inputs = tokenizer(text, return_tensors="pt")
+              It is based on Facebook's <a href="https://about.fb.com/news/2020/10/first-multilingual-machine-translation-model/">M2M-100 model</a>, and can also translate from 100 other languages to Circassian (English, French, Spanish, etc.), but less accurately.
 
-  translated_tokens = model.generate(
-      **inputs, forced_bos_token_id=tokenizer.lang_code_to_id[tgt_lang], num_beams=4, num_return_sequences=4
-  )
+              The data corpus is constantly being expanded, and we need help in finding sentence sources, OCR, data cleaning, etc. 
 
-  translations = []
-  for translation in tokenizer.batch_decode(translated_tokens, skip_special_tokens=True):
-      translations.append(translation)
+              If you are interested in helping out with this project, please contact me at the link below.
+              """
+article = """<p style='text-align: center'><a href='https://arxiv.org/abs/1806.00187'>Scaling Neural Machine Translation</a> | <a href='https://github.com/pytorch/fairseq/'>Github Repo</a></p>"""
 
-  return translations
+examples = [
+  ["Hello world!"],
+  ["PyTorch Hub is a pre-trained model repository designed to facilitate research reproducibility."]
+]
 
-output = gr.outputs.Textbox()
-iface = gr.Interface(fn=translate, inputs="text", outputs=output)
-iface.launch()
+gr.Interface.load("models/anzorq/m2m100_418M_ft_ru-kbd_44K", title=title, description=description, article=article, examples=examples).launch()