Further optimize memory usage: Switch to nllb-200-distilled-200M and add Uvicorn debug logging

- Switched translation model from facebook/nllb-200-distilled-600M to facebook/nllb-200-distilled-200M to reduce memory usage
- Added debug logging around Uvicorn startup to diagnose startup issues
- Added debug print statements in _initialize_tts_model for better logging

app.py

@@ -72,9 +72,10 @@ class TalklasTranslator:
 
     def _initialize_mt_model(self):
         try:
-            self.mt_model = AutoModelForSeq2SeqLM.from_pretrained("facebook/nllb-200-distilled-600M")
+            print("Trying to load facebook/nllb-200-distilled-200M...")
+            self.mt_model = AutoModelForSeq2SeqLM.from_pretrained("facebook/nllb-200-distilled-200M")
             self.mt_tokenizer = AutoTokenizer.from_pretrained(
-                "facebook/nllb-200-distilled-600M",
+                "facebook/nllb-200-distilled-200M",
                 clean_up_tokenization_spaces=True
             )
             self.mt_model.to(self.device)
@@ -84,6 +85,7 @@ class TalklasTranslator:
 
     def _initialize_tts_model(self):
         try:
+            print(f"Trying to load facebook/mms-tts-{self.target_lang}...")
             self.tts_model = VitsModel.from_pretrained(f"facebook/mms-tts-{self.target_lang}")
             self.tts_tokenizer = AutoTokenizer.from_pretrained(
                 f"facebook/mms-tts-{self.target_lang}",
@@ -216,4 +218,6 @@ async def translate_text(text: str = Form(...), source_lang: str = Form(...), ta
 
 if __name__ == "__main__":
     import uvicorn
-    uvicorn.run(app, host="0.0.0.0", port=8000)
+    print("Starting Uvicorn server...")
+    uvicorn.run(app, host="0.0.0.0", port=8000)
+    print("Uvicorn server started successfully")