---
datasets:
- snli
- multi_nli
metrics:
- accuracy
- f1
- precision
- recall
inference: false
model-index:
- name: distilroberta-nli
  results:
  - task:
      type: text-classification
      name: Text Classification
    metrics:
    - type: loss
      value: 0.438475
    - type: accuracy
      value: 0.829536
      name: Accuracy
    - type: f1
      value: 0.828703
      name: F1
    - type: precision
      value: 0.828907
      name: Precision
    - type: recall
      value: 0.828617
      name: Recall
language:
- en
---
## DistilRoBERTa-NLI

This model utilizes the [Distilroberta base](https://huggingface.co/distilroberta-base) architecture, which has been fine-tuned for NLI tasks on the [MultiNLI](https://huggingface.co/datasets/multi_nli) and [SNLI](https://huggingface.co/datasets/snli) datasets. 
It achieves the following results on the evaluation set:
* Loss: 0.4384
* Accuracy: 0.8295


## Model description

The SNLI corpus (version 1.0) is a collection of 570k human-written English sentence pairs manually labeled for balanced classification with the labels entailment, contradiction, and neutral, supporting the task of natural language inference (NLI), also known as recognizing textual entailment (RTE).

The Multi-Genre Natural Language Inference (MultiNLI) corpus is a crowd-sourced collection of 433k sentence pairs annotated with textual entailment information. The corpus is modeled on the SNLI corpus, but differs in that covers a range of genres of spoken and written text, and supports a distinctive cross-genre generalization evaluation.

## Usage
Inference API has been disabled as it is not suitable for this kind of task.

```python
from transformers import AutoTokenizer, AutoModelForSequenceClassification
import torch

# Load model and tokenizer
model_checkpoint = 'AdamCodd/Distilroberta-NLI'
model = AutoModelForSequenceClassification.from_pretrained(model_checkpoint)
tokenizer = AutoTokenizer.from_pretrained(model_checkpoint)

# Set device
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
model.to(device)

# Sample premise and hypothesis
premise = "The cat is sleeping under the sun."
hypothesis = "It's raining, and the cat is getting wet."

# Tokenize and predict
input = tokenizer(premise, hypothesis, truncation=True, padding=True, return_tensors="pt").to(device)
with torch.no_grad():
    output = model(**input)
    probabilities = torch.softmax(output.logits, dim=-1)[0].tolist()

# Output prediction
label_names = ["Entailment", "Neutral", "Contradiction"]
prediction = {name: round(prob * 100, 1) for name, prob in zip(label_names, probabilities)}
print(prediction)
# {'Entailment': 1.3, 'Neutral': 8.2, 'Contradiction': 90.5}
```

## Training and evaluation data

More information needed

## Training procedure

### Training hyperparameters

The following hyperparameters were used during training:
- learning_rate: 3e-05
- train_batch_size: 32
- eval_batch_size: 32
- seed: 42
- optimizer: AdamW with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 150
- num_epochs: 1
- weight_decay: 0.01

### Training results

Metrics: Accuracy, F1, Precision, Recall

```
'eval_loss': 0.438475,
'eval_accuracy': 0.829536,
'eval_f1': 0.828703,
'eval_precision': 0.828907,
'eval_recall': 0.828617
```

### Framework versions

- Transformers 4.36.0
- Datasets 2.15.0
- Tokenizers 0.15.0

If you want to support me, you can [here](https://ko-fi.com/adamcodd).