DeBERTa use for NLI tasks - Missing contradiction score

Hi all,
I was wondering if someone got to make DeBERTa work on NLI task using the pre-trained model available in Hugging Face?

When running this code, the contradiction is always empty (although entailment and neutral score are populated).

from transformers import AutoTokenizer, AutoModelForSequenceClassification
import torch

if __name__ == '__main__':
    max_length = 256

    premise = "Two women are embracing while holding to go packages."
    hypothesis = "The men are fighting outside a deli."

    hg_model_hub_name = "microsoft/deberta-large"
    # hg_model_hub_name = "microsoft/deberta-base"

    tokenizer = AutoTokenizer.from_pretrained(hg_model_hub_name)
    model = AutoModelForSequenceClassification.from_pretrained(hg_model_hub_name)
    tokenized_input_seq_pair = tokenizer.encode_plus(premise, hypothesis,
                                                     return_token_type_ids=True, truncation=True)

    input_ids = torch.Tensor(tokenized_input_seq_pair['input_ids']).long().unsqueeze(0)
    # remember bart doesn't have 'token_type_ids', remove the line below if you are using bart.
    token_type_ids = torch.Tensor(tokenized_input_seq_pair['token_type_ids']).long().unsqueeze(0)
    attention_mask = torch.Tensor(tokenized_input_seq_pair['attention_mask']).long().unsqueeze(0)

    outputs = model(input_ids,
    # Note:
    # "id2label": {
    #     "0": "entailment",
    #     "1": "neutral",
    #     "2": "contradiction"
    # },

    predicted_probability = torch.softmax(outputs[0], dim=1)[0].tolist()  # batch_size only one

    print("Premise:", premise)
    print("Hypothesis:", hypothesis)
    print("Entailment:", predicted_probability[0])
    print("Neutral:", predicted_probability[1])
    print("Contradiction:", predicted_probability[2])
1 Like

The model is not tuned on MNLI task. To inference with MNLI, you need to fine-tuning with MNLI task.