max_length

app.py

@@ -28,7 +28,7 @@ def t5_tranlaste(text:str,model_name:str):
 
     text = f"<2en> {text}"
     input_ids = tokenizer(text, return_tensors="pt").input_ids.to(model.device)
-    outputs = model.generate(input_ids=input_ids, max_new_tokens=1024*2)
+    outputs = model.generate(input_ids=input_ids, max_new_tokens=1024*2,max_length=1024)
 
     translated_text = tokenizer.decode(outputs[0], skip_special_tokens=True)
 
@@ -39,7 +39,7 @@ def mt5_translate(text:str, model_name:str):
     model = MT5ForConditionalGeneration.from_pretrained(models[model_name])
     tokenizer = MT5Tokenizer.from_pretrained(models[model_name])
     inputs = tokenizer(text, return_tensors="pt")
-    result = model.generate(input_ids=inputs['input_ids'], attention_mask=inputs['attention_mask'], max_new_tokens=1024*2)
+    result = model.generate(input_ids=inputs['input_ids'], attention_mask=inputs['attention_mask'], max_new_tokens=1024*2,max_length=1024)
     translated_text = tokenizer.decode(result[0],  skip_special_tokens=True)
     return translated_text