library_name: transformers tags:

• seq2seq license: apache-2.0 datasets:
• Helsinki-NLP/europarl
• Helsinki-NLP/opus-100 language:
• en
• it base_model:
• bigscience/mt0-small pipeline_tag: translation metrics:
• bleu

🍀 四叶草 - 专攻英译意的小型翻译模型

四叶草是基于bigscience/mt0-small架构的编码器-解码器转换器模型，专注于英语-意大利语文本翻译。该模型训练数据来源于Helsinki-NLP/opus-100和Helsinki-NLP/europarl中的英意对照语料。

使用指南

from transformers import AutoTokenizer, AutoModelForSeq2SeqLM

# 从检查点目录加载模型和分词器
tokenizer = AutoTokenizer.from_pretrained("LeonardPuettmann/Quadrifoglio-mt-en-it")
model = AutoModelForSeq2SeqLM.from_pretrained("LeonardPuettmann/Quadrifoglio-mt-en-it")

def generate_response(input_text):
    input_ids = tokenizer("将英语翻译为意大利语:" + input_text, return_tensors="pt").input_ids
    output = model.generate(input_ids, max_new_tokens=256)
    return tokenizer.decode(output[0], skip_special_tokens=True)

text_to_translate = "请给我一杯绿茶。"
response = generate_response(text_to_translate)
print(response)

由于本模型针对单句翻译优化，处理长文本时建议先分句（推荐使用SpaCy），翻译完成后再合并结果：

from transformers import AutoTokenizer, AutoModelForSeq2SeqLM
import spacy
# 首次使用需安装spaCy及英语模型
# !pip install spacy
# !python -m spacy download en_core_web_sm

nlp = spacy.load("en_core_web_sm")

tokenizer = AutoTokenizer.from_pretrained("LeonardPuettmann/Quadrifoglio-mt-en-it")
model = AutoModelForSeq2SeqLM.from_pretrained("LeonardPuettmann/Quadrifoglio-mt-en-it")

def generate_response(input_text):
    input_ids = tokenizer("将意大利语翻译为英语: " + input_text, return_tensors="pt").input_ids
    output = model.generate(input_ids, max_new_tokens=256)
    return tokenizer.decode(output[0], skip_special_tokens=True)

text = "近来可好？今日天气晴好。愿你诸事顺遂。"
doc = nlp(text)
sentences = [sent.text for sent in doc.sents]

sentence_translations = []
for i, sentence in enumerate(sentences):
    sentence_translation = generate_response(sentence)
    sentence_translations.append(sentence_translation)

full_translation = " ".join(sentence_translations)
print(full_translation)

性能评估

基于Opus 100测试集

BLEU指标