srddev/neurohack

This project uses a transformer-based neural network to translate English to Hinglish, a colloquial form of Hindi that incorporates English words and phrases. The model is built from scratch using a large dataset of English-Hindi-Hinglish

NeuroHack

Welcome to NeuroHack, a Machine Learning Hackathon organized by the College of Engineering Pune (COEP). The event is open to students interested in using Machine Learning to solve real-world problems.

Model

This model is available as a Keras model on HuggingFace : model card

from huggingface_hub import from_pretrained_keras

model = from_pretrained_keras("SRDdev/Hin2Hing-Keras_NLP")

Problem Statement

The challenge for this hackathon is to develop a Machine Learning model that can accurately translate English sentences into Hinglish. The dataset provided for training and validation includes English, Hindi, and Hinglish sentences. The test dataset includes only English and Hindi sentences.

Approach

We used Neural Machine Translation using Transformers approach to solve this problem. As there is a lot of similarities in hindi and Hinglish sentence formation, we convert the input english sentence into hindi & then input it to the model which generates a Hinglish sentence.

Dataset

The dataset used for training and validation includes English, Hindi, and Hinglish sentences. The test dataset includes only English and Hindi sentences.

• Dataset

Requirements

To run the code, the following libraries are required:

• TensorFlow
• NumPy
• pandas

Inference

sentence = "Please can you solve this Problem"
output = decode_sequences(sentence)
print(output)
# output : Kripya aap is problem ka samadhan kar sakte he

Contact

In case of any queries or issues, please contact us at [email protected]