Ernie M Large Mnli Xnli

🚀 Multilingual ernie-m-large-mnli-xnli

This multilingual model can perform natural language inference (NLI) on 100 languages and is suitable for multilingual zero - shot classification.

🚀 Quick Start

This multilingual model can perform natural language inference (NLI) on 100 languages, making it well - suited for multilingual zero - shot classification. The underlying model was pre - trained by Baidu, based on Meta's RoBERTa pre - trained on the CC100 multilingual dataset. It was then fine - tuned on the XNLI dataset and the English MNLI dataset.

✨ Features

• Capable of performing natural language inference (NLI) on 100 languages.
• Suitable for multilingual zero - shot classification.
• Outperforms RoBERTa models of equal size.

📦 Installation

The ernie - m architecture is only supported with transformers==4.27 or higher.

• Before the release of 4.27:
  • Install transformers from source: pip install git+https://github.com/huggingface/transformers
  • Install the sentencepiece tokenizer: pip install sentencepiece
• After the release: pip install transformers[sentencepiece]>=4.27

💻 Usage Examples

Basic Usage

from transformers import pipeline
classifier = pipeline("zero-shot-classification", model="MoritzLaurer/ernie-m-large-mnli-xnli")

sequence_to_classify = "Angela Merkel ist eine Politikerin in Deutschland und Vorsitzende der CDU"
candidate_labels = ["politics", "economy", "entertainment", "environment"]
output = classifier(sequence_to_classify, candidate_labels, multi_label=False)
print(output)

Advanced Usage

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

model_name = "MoritzLaurer/ernie-m-large-mnli-xnli"
tokenizer = AutoTokenizer.from_pretrained(model_name)
model = AutoModelForSequenceClassification.from_pretrained(model_name).to(device)

premise = "Angela Merkel ist eine Politikerin in Deutschland und Vorsitzende der CDU"
hypothesis = "Emmanuel Macron is the President of France"

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)

📚 Documentation

Training data

This model was trained on the XNLI development dataset and the MNLI train dataset. The XNLI development set consists of 2490 professionally translated texts from English to 14 other languages (37350 texts in total). Not using machine - translated texts can avoid overfitting the model to the 15 languages, avoid catastrophic forgetting of the other 85 languages ernie - m was pre - trained on, and significantly reduce training costs.

Training procedure

ernie - m - large - mnli - xnli was trained using the Hugging Face trainer with the following hyperparameters.

training_args = TrainingArguments(
    num_train_epochs=3,              # total number of training epochs
    learning_rate=3e-05,
    per_device_train_batch_size=16,   # batch size per device during training
    gradient_accumulation_steps=2,
    per_device_eval_batch_size=16,    # batch size for evaluation
    warmup_ratio=0.1,                # number of warmup steps for learning rate scheduler
    weight_decay=0.01,               # strength of weight decay
    fp16=True,
)

Eval results

The model was evaluated on the XNLI test set on 15 languages (5010 texts per language, 75150 in total). Multilingual NLI models can classify NLI texts without receiving NLI training data in the specific language (cross - lingual transfer).

🔧 Technical Details

The underlying model was pre - trained by Baidu, based on Meta's RoBERTa pre - trained on the CC100 multilingual dataset. It was then fine - tuned on the XNLI dataset and the English MNLI dataset. The model was introduced by Baidu in this paper.

📄 License

This model is licensed under the apache - 2.0 license.

Limitations and bias

Please consult the original ernie - m paper and literature on different NLI datasets for potential biases.

Citation

If you use this model, please cite: Laurer, Moritz, Wouter van Atteveldt, Andreu Salleras Casas, and 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.

Ideas for cooperation or questions?

If you have questions or ideas for cooperation, contact me at m{dot}laurer{at}vu{dot}nl or [LinkedIn](https://www.linkedin.com/in/moritz - laurer/)

Debugging and issues

⚠️ Important Note

The ernie - m architecture is only supported with transformers==4.27 or higher (which is not yet released and causes an error in the inference widget as of 03.03.23).

💡 Usage Tip

In order to run the model before the release of 4.27, you need to install transformers from source with: pip install git+https://github.com/huggingface/transformers as well as the sentencepiece tokenizer with: pip install sentencepiece. After the release, you can run: pip install transformers[sentencepiece]>=4.27