模型:
setu4993/LEALLA-small
LEALLA是从 LaBSE 个训练模型中精简而来的一套支持109种语言的轻量级语言无关句子嵌入模型。该模型可用于获取多语言句子嵌入以及双语文本检索。
此模型从TF Hub的v1模型迁移而来。两个版本的模型产生的嵌入结果均为 equivalent 。不过,对于某些语言(如日语),LEALLA模型在比较嵌入和相似度时似乎需要更高的容差。
使用该模型:
import torch from transformers import BertModel, BertTokenizerFast tokenizer = BertTokenizerFast.from_pretrained("setu4993/LEALLA-small") model = BertModel.from_pretrained("setu4993/LEALLA-small") model = model.eval() english_sentences = [ "dog", "Puppies are nice.", "I enjoy taking long walks along the beach with my dog.", ] english_inputs = tokenizer(english_sentences, return_tensors="pt", padding=True) with torch.no_grad(): english_outputs = model(**english_inputs)
要获得句子嵌入,请使用pooler输出:
english_embeddings = english_outputs.pooler_output
其他语言的输出:
italian_sentences = [ "cane", "I cuccioli sono carini.", "Mi piace fare lunghe passeggiate lungo la spiaggia con il mio cane.", ] japanese_sentences = ["犬", "子犬はいいです", "私は犬と一緒にビーチを散歩するのが好きです"] italian_inputs = tokenizer(italian_sentences, return_tensors="pt", padding=True) japanese_inputs = tokenizer(japanese_sentences, return_tensors="pt", padding=True) with torch.no_grad(): italian_outputs = model(**italian_inputs) japanese_outputs = model(**japanese_inputs) italian_embeddings = italian_outputs.pooler_output japanese_embeddings = japanese_outputs.pooler_output
对于句子之间的相似度计算,建议在计算相似度之前进行L2范数归一化:
import torch.nn.functional as F def similarity(embeddings_1, embeddings_2): normalized_embeddings_1 = F.normalize(embeddings_1, p=2) normalized_embeddings_2 = F.normalize(embeddings_2, p=2) return torch.matmul( normalized_embeddings_1, normalized_embeddings_2.transpose(0, 1) ) print(similarity(english_embeddings, italian_embeddings)) print(similarity(english_embeddings, japanese_embeddings)) print(similarity(italian_embeddings, japanese_embeddings))
关于数据、训练、评估和性能指标的详细信息,请参阅 original paper 。
@misc{mao2023lealla, title={LEALLA: Learning Lightweight Language-agnostic Sentence Embeddings with Knowledge Distillation}, author={Zhuoyuan Mao and Tetsuji Nakagawa}, year={2023}, eprint={2302.08387}, archivePrefix={arXiv}, primaryClass={cs.CL} }