DeBERTa-v3-base-mnli-fever-anli

模型描述

该模型是在MultiNLI、Fever-NLI和Adversarial-NLI（ANLI）数据集上进行训练的，包括763,913个NLI假设-前提对。该基础模型在 ANLI benchmark 上优于几乎所有大型模型。基础模型是 DeBERTa-v3-base from Microsoft 。DeBERTa的v3变体通过包含不同的预训练目标显著优于模型的先前版本，请参阅原始 DeBERTa paper 的附录11。

为了最高效的性能（但速度较慢），建议使用 https://huggingface.co/MoritzLaurer/DeBERTa-v3-large-mnli-fever-anli-ling-wanli 。

如何使用该模型

from transformers import pipeline
classifier = pipeline("zero-shot-classification", model="MoritzLaurer/DeBERTa-v3-base-mnli-fever-anli")
sequence_to_classify = "Angela Merkel is a politician in Germany and leader of the CDU"
candidate_labels = ["politics", "economy", "entertainment", "environment"]
output = classifier(sequence_to_classify, candidate_labels, multi_label=False)
print(output)

from transformers import AutoTokenizer, AutoModelForSequenceClassification
import torch
device = torch.device("cuda") if torch.cuda.is_available() else torch.device("cpu")

model_name = "MoritzLaurer/DeBERTa-v3-base-mnli-fever-anli"
tokenizer = AutoTokenizer.from_pretrained(model_name)
model = AutoModelForSequenceClassification.from_pretrained(model_name)

premise = "I first thought that I liked the movie, but upon second thought it was actually disappointing."
hypothesis = "The movie was good."

input = tokenizer(premise, hypothesis, truncation=True, return_tensors="pt")
output = model(input["input_ids"].to(device))  # device = "cuda:0" or "cpu"
prediction = torch.softmax(output["logits"][0], -1).tolist()
label_names = ["entailment", "neutral", "contradiction"]
prediction = {name: round(float(pred) * 100, 1) for pred, name in zip(prediction, label_names)}
print(prediction)

训练数据

DeBERTa-v3-base-mnli-fever-anli是在MultiNLI、Fever-NLI和Adversarial-NLI（ANLI）数据集上进行训练的，包括763,913个NLI假设-前提对。

训练过程

DeBERTa-v3-base-mnli-fever-anli使用Hugging Face的trainer训练，采用以下超参数。

training_args = TrainingArguments(
    num_train_epochs=3,              # total number of training epochs
    learning_rate=2e-05,
    per_device_train_batch_size=32,   # batch size per device during training
    per_device_eval_batch_size=32,    # batch size for evaluation
    warmup_ratio=0.1,                # number of warmup steps for learning rate scheduler
    weight_decay=0.06,               # strength of weight decay
    fp16=True                        # mixed precision training
)

评估结果

模型使用MultiNLI的测试集、ANLI的测试集和Fever-NLI的开发集进行评估，使用的度量是准确率。

限制和偏见

请参考原始的DeBERTa论文和不同的NLI数据集的文献，了解潜在的偏见。

引用

如果您使用该模型，请引用：Laurer, Moritz, Wouter van Atteveldt, Andreu Salleras Casas和Kasper Welbers. 2022.‘Less Annotating, More Classifying – Addressing the Data Scarcity Issue of Supervised Machine Learning with Deep Transfer Learning and BERT - NLI’. Preprint, June. Open Science Framework. https://osf.io/74b8k 。

合作或问题的想法?

如果您有问题或合作的想法，请通过m{点}laurer{at}vu{点}nl或 LinkedIn 与我联系。

调试和问题

请注意，DeBERTa-v3发布于06.12.21，旧版本的HF Transformers似乎在运行该模型时存在问题（例如，与标记器相关的问题）。使用Transformers>=4.13可能会解决一些问题。

模型回收

Evaluation on 36 datasets 使用MoritzLaurer/DeBERTa-v3-base-mnli-fever-anli作为基础模型，与microsoft/deberta-v3-base相比，平均得分为79.69，而microsoft/deberta-v3-base为79.04。

该模型在截止到2023年01月09日的所有测试模型中排名第二，针对microsoft/deberta-v3-base架构。

结果：

更多信息，请参阅： Model Recycling