# DeBERTa-v3-small-mnli-fever-docnli-ling-2c

## Model description

This model was trained on 1.279.665 hypothesis-premise pairs from 8 NLI datasets: MultiNLI, Fever-NLI, LingNLI and DocNLI (which includes ANLI, QNLI, DUC, CNN/DailyMail, Curation).

It is the only model in the model hub trained on 8 NLI datasets, including DocNLI with very long texts to learn long range reasoning. Note that the model was trained on binary NLI to predict either "entailment" or "not-entailment". The DocNLI merges the classes "neural" and "contradiction" into "not-entailment" to create more training data.

The base model is DeBERTa-v3-small from Microsoft. The v3 variant of DeBERTa substantially outperforms previous versions of the model by including a different pre-training objective, see annex 11 of the original DeBERTa paper as well as the DeBERTa-V3 paper.

## Intended uses & limitations

#### How to use the model

from transformers import AutoTokenizer, AutoModelForSequenceClassification
import torch

model_name = "MoritzLaurer/DeBERTa-v3-small-mnli-fever-docnli-ling-2c"
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()
prediction = {name: round(float(pred) * 100, 1) for pred, name in zip(prediction, label_names)}
print(prediction)


### Training data

This model was trained on 1.279.665 hypothesis-premise pairs from 8 NLI datasets: MultiNLI, Fever-NLI, LingNLI and DocNLI (which includes ANLI, QNLI, DUC, CNN/DailyMail, Curation).

### Training procedure

DeBERTa-v3-small-mnli-fever-docnli-ling-2c 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=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
)


### Eval results

The model was evaluated using the binary test sets for MultiNLI and ANLI and the binary dev set for Fever-NLI (two classes instead of three). The metric used is accuracy.

mnli-m-2c mnli-mm-2c fever-nli-2c anli-all-2c anli-r3-2c
0.927 0.921 0.892 0.684 0.673

## Limitations and bias

Please consult the original DeBERTa paper and literature on different NLI datasets for potential biases.

### BibTeX entry and citation info

If you want to cite this model, please cite the original DeBERTa paper, the respective NLI datasets and include a link to this model on the Hugging Face hub.

### 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

### Debugging and issues

Note that DeBERTa-v3 was released recently and older versions of HF Transformers seem to have issues running the model (e.g. resulting in an issue with the tokenizer). Using Transformers==4.13 might solve some issues.

Mask token: undefined