NorBERT 3 base

新一代NorBERT语言模型的官方发布，论文[ NorBench — A Benchmark for Norwegian Language Models ]中详细描述了该模型的更多细节。请阅读论文以了解更多详情。

其他规模：

• NorBERT 3 xs (15M)
• NorBERT 3 small (40M)
• NorBERT 3 base (123M)
• NorBERT 3 large (323M)

生成型NorT5衍生模型：

• NorT5 xs (32M)
• NorT5 small (88M)
• NorT5 base (228M)
• NorT5 large (808M)

示例用法：

当前此模型需要使用从"modeling_norbert.py"中加载的自定义wrapper，因此应以trust_remote_code=True加载模型。

import torch
from transformers import AutoTokenizer, AutoModelForMaskedLM

tokenizer = AutoTokenizer.from_pretrained("ltg/norbert3-base")
model = AutoModelForMaskedLM.from_pretrained("ltg/norbert3-base", trust_remote_code=True)

mask_id = tokenizer.convert_tokens_to_ids("[MASK]")
input_text = tokenizer("Nå ønsker de seg en[MASK] bolig.", return_tensors="pt")
output_p = model(**input_text)
output_text = torch.where(input_text.input_ids == mask_id, output_p.logits.argmax(-1), input_text.input_ids)

# should output: '[CLS] Nå ønsker de seg en ny bolig.[SEP]'
print(tokenizer.decode(output_text[0].tolist()))

当前已实现以下类：AutoModel，AutoModelMaskedLM，AutoModelForSequenceClassification，AutoModelForTokenClassification，AutoModelForQuestionAnswering和AutoModeltForMultipleChoice。

引用我们

@inproceedings{samuel-etal-2023-norbench,
    title = "{N}or{B}ench {--} A Benchmark for {N}orwegian Language Models",
    author = "Samuel, David  and
      Kutuzov, Andrey  and
      Touileb, Samia  and
      Velldal, Erik  and
      {\O}vrelid, Lilja  and
      R{\o}nningstad, Egil  and
      Sigdel, Elina  and
      Palatkina, Anna",
    booktitle = "Proceedings of the 24th Nordic Conference on Computational Linguistics (NoDaLiDa)",
    month = may,
    year = "2023",
    address = "T{\'o}rshavn, Faroe Islands",
    publisher = "University of Tartu Library",
    url = "https://aclanthology.org/2023.nodalida-1.61",
    pages = "618--633",
    abstract = "We present NorBench: a streamlined suite of NLP tasks and probes for evaluating Norwegian language models (LMs) on standardized data splits and evaluation metrics. We also introduce a range of new Norwegian language models (both encoder and encoder-decoder based). Finally, we compare and analyze their performance, along with other existing LMs, across the different benchmark tests of NorBench.",
}