language:
- en
license: mit
tags:
- text-classification
- zero-shot-classification
metrics:
- accuracy
datasets:
- multi_nli
- anli
- fever
pipeline_tag: zero-shot-classification
model-index:
- name: MoritzLaurer/DeBERTa-v3-base-mnli-fever-anli
results:
- task:
type: natural-language-inference
name: Natural Language Inference
dataset:
name: anli
type: anli
config: plain_text
split: test_r3
metrics:
- name: Accuracy
type: accuracy
value: 0.495
verified: true
- name: Precision Macro
type: precision
value: 0.4984740618243923
verified: true
- name: Precision Micro
type: precision
value: 0.495
verified: true
- name: Precision Weighted
type: precision
value: 0.4984357572868885
verified: true
- name: Recall Macro
type: recall
value: 0.49461028192371476
verified: true
- name: Recall Micro
type: recall
value: 0.495
verified: true
- name: Recall Weighted
type: recall
value: 0.495
verified: true
- name: F1 Macro
type: f1
value: 0.4942810999491704
verified: true
- name: F1 Micro
type: f1
value: 0.495
verified: true
- name: F1 Weighted
type: f1
value: 0.4944671868893595
verified: true
- name: loss
type: loss
value: 1.8788293600082397
verified: true
- task:
type: natural-language-inference
name: Natural Language Inference
dataset:
name: anli
type: anli
config: plain_text
split: test_r1
metrics:
- name: Accuracy
type: accuracy
value: 0.712
verified: true
- name: Precision Macro
type: precision
value: 0.7134839439315348
verified: true
- name: Precision Micro
type: precision
value: 0.712
verified: true
- name: Precision Weighted
type: precision
value: 0.7134676028447461
verified: true
- name: Recall Macro
type: recall
value: 0.7119814425203647
verified: true
- name: Recall Micro
type: recall
value: 0.712
verified: true
- name: Recall Weighted
type: recall
value: 0.712
verified: true
- name: F1 Macro
type: f1
value: 0.7119226991285647
verified: true
- name: F1 Micro
type: f1
value: 0.712
verified: true
- name: F1 Weighted
type: f1
value: 0.7119242267218338
verified: true
- name: loss
type: loss
value: 1.0105403661727905
verified: true
- task:
type: natural-language-inference
name: Natural Language Inference
dataset:
name: multi_nli
type: multi_nli
config: default
split: validation_mismatched
metrics:
- name: Accuracy
type: accuracy
value: 0.902766476810415
verified: true
- name: Precision Macro
type: precision
value: 0.9023816542652491
verified: true
- name: Precision Micro
type: precision
value: 0.902766476810415
verified: true
- name: Precision Weighted
type: precision
value: 0.9034597464719761
verified: true
- name: Recall Macro
type: recall
value: 0.9024304801555488
verified: true
- name: Recall Micro
type: recall
value: 0.902766476810415
verified: true
- name: Recall Weighted
type: recall
value: 0.902766476810415
verified: true
- name: F1 Macro
type: f1
value: 0.9023086094638595
verified: true
- name: F1 Micro
type: f1
value: 0.902766476810415
verified: true
- name: F1 Weighted
type: f1
value: 0.9030161011457231
verified: true
- name: loss
type: loss
value: 0.3283354640007019
verified: true
DeBERTa-v3-base-mnli-fever-anli
Model description
This model was trained on the MultiNLI, Fever-NLI and Adversarial-NLI (ANLI) datasets, which comprise 763 913 NLI hypothesis-premise pairs. This base model outperforms almost all large models on the ANLI benchmark. 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 highest performance (but less speed), I recommend using https://huggingface.co/MoritzLaurer/DeBERTa-v3-large-mnli-fever-anli-ling-wanli.
How to use the model
Simple zero-shot classification pipeline
from transformers import pipeline
classifier = pipeline("zero-shot-classification", model="MoritzLaurer/mDeBERTa-v3-base-mnli-xnli")
sequence_to_classify = "Angela Merkel is a politician in Germany and leader of the CDU 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)
NLI use-case
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)
Training data
DeBERTa-v3-base-mnli-fever-anli was trained on the MultiNLI, Fever-NLI and Adversarial-NLI (ANLI) datasets, which comprise 763 913 NLI hypothesis-premise pairs.
Training procedure
DeBERTa-v3-base-mnli-fever-anli 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 test sets for MultiNLI and ANLI and the dev set for Fever-NLI. The metric used is accuracy.
| mnli-m | mnli-mm | fever-nli | anli-all | anli-r3 |
|---|---|---|---|---|
| 0.903 | 0.903 | 0.777 | 0.579 | 0.495 |
Limitations and bias
Please consult the original DeBERTa 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
Debugging and issues
Note that DeBERTa-v3 was released on 06.12.21 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.