TiberiuCristianLeon commited on
Commit
153e2d0
·
verified ·
1 Parent(s): 9f3478a

Update app.py

Browse files
Files changed (1) hide show
  1. app.py +4 -3
app.py CHANGED
@@ -207,11 +207,12 @@ def load_diacritics_model():
207
  @app.get("/diacritics", operation_id="convert_diacritics", description="Romanian Diacritics Restoration", tags=["diacritics"], summary="Romanian diacritics")
208
  def diacritics(text: str, model: str = DIACRITICS_MODEL):
209
  model, tokenizer = load_diacritics_model()
210
- inputs = tokenizer(text, max_length=512, truncation=True, return_tensors="pt")
211
- outputs = model.generate(input_ids=inputs["input_ids"], attention_mask=inputs["attention_mask"])
 
212
  output = tokenizer.decode(outputs[0], skip_special_tokens=True)
213
  info = load_diacritics_model.cache_info() # check if loaded from cache
214
- return {"input": text, "diacritics": output, "length": (len(output), info.hits)}
215
  return output
216
 
217
  # Create an MCP server based on this app
 
207
  @app.get("/diacritics", operation_id="convert_diacritics", description="Romanian Diacritics Restoration", tags=["diacritics"], summary="Romanian diacritics")
208
  def diacritics(text: str, model: str = DIACRITICS_MODEL):
209
  model, tokenizer = load_diacritics_model()
210
+ inputs = tokenizer(text, max_length=256, truncation=True, return_tensors="pt")
211
+ outputs = model.generate(input_ids=inputs["input_ids"], attention_mask=inputs["attention_mask"],
212
+ max_new_tokens=256, num_beams=3, early_stopping=True)
213
  output = tokenizer.decode(outputs[0], skip_special_tokens=True)
214
  info = load_diacritics_model.cache_info() # check if loaded from cache
215
+ return {"input": text, "diacritics": output, "Info": (len(output), info.hits)}
216
  return output
217
 
218
  # Create an MCP server based on this app