模型:
camembert-base
该模型可用于填充掩码任务。
内容警告:读者应该知道,本部分包含令人不安、冒犯性和能够传播历史和当前刻板印象的内容。
已进行了大量研究,探讨了语言模型的偏见和公平性问题(参见,例如, Sheng et al. (2021) 和 Bender et al. (2021) )。
该模型在OSCAR多语言语料库的一个子语料库上进行了预训练。与OSCAR数据集相关的一些限制和风险在 OSCAR dataset card 中详细说明,包括:
某些OSCAR子语料库的质量可能低于预期,特别是对于资源最低的语言。
构建自Common Crawl,可能存在个人和敏感信息。
OSCAR或Open Super-large Crawled Aggregated coRpus是通过使用Ungoliant架构对Common Crawl语料库进行语言分类和过滤而获得的多语种语料库。
训练过程Model | #params | Arch. | Training data |
---|---|---|---|
camembert-base | 110M | Base | OSCAR (138 GB of text) |
camembert/camembert-large | 335M | Large | CCNet (135 GB of text) |
camembert/camembert-base-ccnet | 110M | Base | CCNet (135 GB of text) |
camembert/camembert-base-wikipedia-4gb | 110M | Base | Wikipedia (4 GB of text) |
camembert/camembert-base-oscar-4gb | 110M | Base | Subsample of OSCAR (4 GB of text) |
camembert/camembert-base-ccnet-4gb | 110M | Base | Subsample of CCNet (4 GB of text) |
模型开发者使用四种不同的法语下游任务对CamemBERT进行了评估:词性标注(POS)、依存句法分析、命名实体识别(NER)和自然语言推理(NLI)。
@inproceedings{martin2020camembert, title={CamemBERT: a Tasty French Language Model}, author={Martin, Louis and Muller, Benjamin and Su{\'a}rez, Pedro Javier Ortiz and Dupont, Yoann and Romary, Laurent and de la Clergerie, {\'E}ric Villemonte and Seddah, Djam{\'e} and Sagot, Beno{\^\i}t}, booktitle={Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics}, year={2020} }
from transformers import CamembertModel, CamembertTokenizer # You can replace "camembert-base" with any other model from the table, e.g. "camembert/camembert-large". tokenizer = CamembertTokenizer.from_pretrained("camembert-base") camembert = CamembertModel.from_pretrained("camembert-base") camembert.eval() # disable dropout (or leave in train mode to finetune)使用管道方法填充掩码:
from transformers import pipeline camembert_fill_mask = pipeline("fill-mask", model="camembert-base", tokenizer="camembert-base") results = camembert_fill_mask("Le camembert est <mask> :)") # results #[{'sequence': '<s> Le camembert est délicieux :)</s>', 'score': 0.4909103214740753, 'token': 7200}, # {'sequence': '<s> Le camembert est excellent :)</s>', 'score': 0.10556930303573608, 'token': 2183}, # {'sequence': '<s> Le camembert est succulent :)</s>', 'score': 0.03453315049409866, 'token': 26202}, # {'sequence': '<s> Le camembert est meilleur :)</s>', 'score': 0.03303130343556404, 'token': 528}, # {'sequence': '<s> Le camembert est parfait :)</s>', 'score': 0.030076518654823303, 'token': 1654}]从CamemBERT输出中提取上下文嵌入特征:
import torch # Tokenize in sub-words with SentencePiece tokenized_sentence = tokenizer.tokenize("J'aime le camembert !") # ['▁J', "'", 'aime', '▁le', '▁ca', 'member', 't', '▁!'] # 1-hot encode and add special starting and end tokens encoded_sentence = tokenizer.encode(tokenized_sentence) # [5, 121, 11, 660, 16, 730, 25543, 110, 83, 6] # NB: Can be done in one step : tokenize.encode("J'aime le camembert !") # Feed tokens to Camembert as a torch tensor (batch dim 1) encoded_sentence = torch.tensor(encoded_sentence).unsqueeze(0) embeddings, _ = camembert(encoded_sentence) # embeddings.detach() # embeddings.size torch.Size([1, 10, 768]) # tensor([[[-0.0254, 0.0235, 0.1027, ..., -0.1459, -0.0205, -0.0116], # [ 0.0606, -0.1811, -0.0418, ..., -0.1815, 0.0880, -0.0766], # [-0.1561, -0.1127, 0.2687, ..., -0.0648, 0.0249, 0.0446], # ...,从所有CamemBERT层中提取上下文嵌入特征:
from transformers import CamembertConfig # (Need to reload the model with new config) config = CamembertConfig.from_pretrained("camembert-base", output_hidden_states=True) camembert = CamembertModel.from_pretrained("camembert-base", config=config) embeddings, _, all_layer_embeddings = camembert(encoded_sentence) # all_layer_embeddings list of len(all_layer_embeddings) == 13 (input embedding layer + 12 self attention layers) all_layer_embeddings[5] # layer 5 contextual embedding : size torch.Size([1, 10, 768]) #tensor([[[-0.0032, 0.0075, 0.0040, ..., -0.0025, -0.0178, -0.0210], # [-0.0996, -0.1474, 0.1057, ..., -0.0278, 0.1690, -0.2982], # [ 0.0557, -0.0588, 0.0547, ..., -0.0726, -0.0867, 0.0699], # ...,