Python 如何使用变形金刚加快翻译速度？

我试着把一个数据框从英语翻译成波斯语。为此，我使用了一个持续训练的朗格模型，但它很慢，我如何才能加速它呢

model_size = "base"
model_name = f"persiannlp/mt5-{model_size}-parsinlu-translation_en_fa"
tokenizer = MT5Tokenizer.from_pretrained(model_name)
model = MT5ForConditionalGeneration.from_pretrained(model_name)



with open(fname, 'w') as tsvfile:
    writer = csv.writer(tsvfile, delimiter='\t')
    #writer.writerow(["prefix", "input_text", "target_text"])    
    writer.writerow(["input_text", "target_text"])    
    for i, row in tqdm(df.iterrows(), total=maxlen): 
        prompt = row["input_text"]
        target = row["target_text"]
        if trans:
            input_ids = tokenizer.encode(prompt, return_tensors="pt")
            res = model.generate(input_ids)
            prompt = tokenizer.batch_decode(res, skip_special_tokens=True)
            input_ids = tokenizer.encode(target, return_tensors="pt")
            res = model.generate(input_ids)
            target = tokenizer.batch_decode(res, skip_special_tokens=True)
        writer.writerow([prompt, target])                

print("saved in ", fname)        

---

这回答了你的问题吗@谢谢，我检查过了，它缺少英语到波斯语的模式（en-fa）。我对tensorflow不太熟悉，但我可能可以将数据集转换为tensorflow格式，然后在其上调用模型或类似的解决方案。@Jindřich