kamalchaurasia-iitj/mlops-anli-classifier-roberta

ANLI NLI Classifier (RoBERTa)

Fine-tuned roberta-base for 3-class natural language inference on the ANLI dataset.

Labels

• entailment
• neutral
• contradiction

Training details

Parameter	Value
Base model	roberta-base
Epochs	4
Learning rate	3e-05
Train batch size	16
Warmup steps	150
Weight decay	0.01

Evaluation (test set)

Metric	Score
Accuracy	0.4400
Weighted F1	0.4304
Loss	2.8117

Usage

from transformers import pipeline

classifier = pipeline(
    "text-classification",
    model="kamalchaurasia-iitj/mlops-anli-classifier-roberta",
    tokenizer="kamalchaurasia-iitj/mlops-anli-classifier-roberta",
)

text = "premise: A man is playing guitar on stage. hypothesis: A person is performing music."
result = classifier(text)
print(result)

Citation

@misc{anli-nli-classifier,
  author = {Kamalkumar Chaurasia},
  title  = {ANLI NLI Classifier (roberta-base)},
  year   = {2026},
  url    = {https://huggingface.co/kamalchaurasia-iitj/mlops-anli-classifier-roberta}
}