mobilebert uncased mnli

Introduction

MobileBERT-uncased-MNLI is a fine-tuned version of the uncased MobileBERT model designed for zero-shot classification tasks. Developed and shared by Typeform, it operates primarily in the English language.

Architecture

MobileBERT-uncased-MNLI is a compact, task-agnostic version of BERT, suitable for resource-limited devices. Its architecture focuses on delivering efficient performance for zero-shot classification.

Training

The model is fine-tuned using the Multi-Genre Natural Language Inference (MNLI) dataset. It is designed to handle various natural language understanding tasks without requiring extensive computational resources. For detailed training data, refer to the multi_nli dataset card.

Guide: Running Locally

To use MobileBERT-uncased-MNLI locally, follow these steps:

1. Install the Transformers Library:

   pip install transformers
   

2. Load the Tokenizer and Model:

   from transformers import AutoTokenizer, AutoModelForSequenceClassification
   
   tokenizer = AutoTokenizer.from_pretrained("typeform/mobilebert-uncased-mnli")
   model = AutoModelForSequenceClassification.from_pretrained("typeform/mobilebert-uncased-mnli")
   

3. Run Inference: Utilize the tokenizer and model to process and classify your text data.

For optimal performance, especially with larger datasets, consider using cloud-based GPU services such as AWS, Google Cloud, or Azure.

License

The licensing information for MobileBERT-uncased-MNLI is not explicitly provided. Users should verify the licensing terms before using the model in production environments.