You can deploy this model and start earning money today!

startificial/
nli-implementation

Name: nli-implementation
Rating: 2.5 (13 reviews)

zero-shot-classification

2006.03654

endpoints_compatible

Views

No views yet

This specific model builds on the repo MoritzLaurer/DeBERTa-v3-base-mnli and adds a handler.py file for more easily utilizing the model in inference endpoints to make zero shot classification comparisons of premise and hypothesis as entailment, neutral, or contradiction.

DeBERTa-v3-base-mnli-fever-anli

Model description

This model was trained on the MultiNLI dataset, which consists of 392 702 NLI hypothesis-premise pairs. The base model is DeBERTa-v3-base 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. For a more powerful model, check out DeBERTa-v3-base-mnli-fever-anli which was trained on even more data.

Intended uses & limitations

How to use the model

python

from transformers import AutoTokenizer, AutoModelForSequenceClassification
import torch
model_name = "MoritzLaurer/DeBERTa-v3-base-mnli"
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)

Example cURL:

curl YOUR INFERENCE ENDPOINT URL HERE \ -X POST \ -d '{"inputs": {"premise": "A man is walking his dog in the park.", "hypothesis": "A person is outside with an animal."}}' \ -H "Authorization: Bearer hf_YOUR_TOKEN_HERE" \ -H "Content-Type: application/json

Training data

This model was trained on the MultiNLI dataset, which consists of 392 702 NLI hypothesis-premise pairs.

Training procedure

DeBERTa-v3-base-mnli was trained using the Hugging Face trainer with the following hyperparameters.

training_args = TrainingArguments(
    num_train_epochs=5,              # 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 matched test set and achieves 0.90 accuracy.

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.

Model Recycling

Evaluation on 36 datasets using MoritzLaurer/DeBERTa-v3-base-mnli as a base model yields average score of 80.01 in comparison to 79.04 by microsoft/deberta-v3-base.

The model is ranked 1st among all tested models for the microsoft/deberta-v3-base architecture as of 09/01/2023.

Results:

20_newsgroup	ag_news	amazon_reviews_multi	anli	boolq	cb	cola	copa	dbpedia	esnli	financial_phrasebank	imdb	isear	mnli	mrpc	multirc	poem_sentiment	qnli	qqp	rotten_tomatoes	rte	sst2	sst_5bins	stsb	trec_coarse	trec_fine	tweet_ev_emoji	tweet_ev_emotion	tweet_ev_hate	tweet_ev_irony	tweet_ev_offensive	tweet_ev_sentiment	wic	wnli	wsc	yahoo_answers
86.0196	90.6333	66.96	60.0938	83.792	83.9286	86.5772	72	79.2	91.419	85.1	94.232	71.5124	89.4426	90.4412	63.7583	86.5385	93.8129	91.9144	89.8687	85.9206	95.4128	57.3756	91.377	97.4	91	47.302	83.6031	57.6431	77.1684	83.3721	70.2947	71.7868	67.6056	74.0385	71.7

For more information, see: Model Recycling

startificial/nli-implementation