Giter VIP home page Giter VIP logo

hinglish's Introduction

Hinglish

Tools and Data

Data and Model files

Logo

Approach LM Perplexity Classifier F1
BERT 8.2 0.63
DistilBERT 6.5 0.63
ULMFIT 21 0.61
RoBERTa 7.54 0.64

Model Performance

Base LM Dataset Accuracy Precision Recall F1 LM Perplexity
bert-base-multilingual-cased Test 0.688 0.698 0.686 0.687 8.2
bert-base-multilingual-cased Valid 0.62 0.592 0.605 0.55 8.2
distilbert-base-uncased Test 0.693 0.694 0.703 0.698 6.51
distilbert-base-uncased Valid 0.607 0.614 0.600 0.592 6.51
distilbert-base-multilingual-cased Test 0.612 0.615 0.616 0.616 8.1
distilbert-base-multilingual-cased Valid 0.55 0.531 0.537 0.495 8.1
roberta-base Test 0.630 0.629 0.644 0.635 7.54
roberta-base Valid 0.60 0.617 0.607 0.595 7.54
Ensemble Test 0.714 0.718 0.718 0.718

Ensemble Performace

Model Accuracy Precision Recall F1 Config Link to Model and output files
BERT 0.68866 0.69821 0.68608 0.6875 Batch Size - 16
Attention Dropout - 0.4
Learning Rate - 5e-07
Adam epsilon - 1e-08
Hidden Dropout Probability - 0.3
Epochs - 3		 BERT
DistilBert 0.69333 0.69496 0.70379 0.6982 Batch Size - 16
Attention Dropout - 0.6
Learning Rate - 3e-05
Adam epsilon - 1e-08
Hidden Dropout Probability - 0.6
Epochs - 3		 DistilBert
EnsembleBert1 0.69233 0.70236 0.69064 0.68952 Batch Size - 4
Attention Dropout - 0.7
Learning Rate - 5.01e-05
Adam epsilon - 4.79e-05
Hidden Dropout Probability - 0.1
Epochs - 3		 EnsembleBert1
EnsembleBert2 0.691 0.7009 0.6889 0.68872 Batch Size - 4
Attention Dropout - 0.6
Learning Rate - 5.13e-05
Adam epsilon - 9.72e-05
Hidden Dropout Probability - 0.2
Epochs - 3		 EnsembleBert2
EnsembleDistilBert1 0.70166 0.70377 0.70976 0.7061 Batch Size - 16
Attention Dropout - 0.8
Learning Rate - 3.02e-05
Adam epsilon - 9.35e-05
Hidden Dropout Probability - 0.4
Epochs - 3		 EnsembleDistilBert1
EnsembleDistilBert2 0.689 0.691 0.69666 0.69335 Batch Size - 4
Attention Dropout - 0.6
Learning Rate - 5.13e-05
Adam epsilon - 9.72e-05
Hidden Dropout Probability - 0.2
Epochs - 3		 EnsembleDistilBert2
EnsembleDistilBert3 0.69366 0.69538 0.70557 0.69905 Batch Size - 16
Attention Dropout - 0.4
Learning Rate - 4.74e-05
Adam epsilon - 4.09e-05
Hidden Dropout Probability - 0.6
Epochs - 3		 EnsembleDistilBert3
Ensemble 0.71466 0.71867 0.71853 0.7182 NA Ensemble

hinglish's People

Contributors

meghanabhange avatar nirantk avatar

Watchers

avatar

Recommend Projects

  • React photo React

    A declarative, efficient, and flexible JavaScript library for building user interfaces.

  • Vue.js photo Vue.js

    🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.

  • Typescript photo Typescript

    TypeScript is a superset of JavaScript that compiles to clean JavaScript output.

  • TensorFlow photo TensorFlow

    An Open Source Machine Learning Framework for Everyone

  • Django photo Django

    The Web framework for perfectionists with deadlines.

  • D3 photo D3

    Bring data to life with SVG, Canvas and HTML. 📊📈🎉

Recommend Topics

  • javascript

    JavaScript (JS) is a lightweight interpreted programming language with first-class functions.

  • web

    Some thing interesting about web. New door for the world.

  • server

    A server is a program made to process requests and deliver data to clients.

  • Machine learning

    Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.

  • Game

    Some thing interesting about game, make everyone happy.

Recommend Org

  • Facebook photo Facebook

    We are working to build community through open source technology. NB: members must have two-factor auth.

  • Microsoft photo Microsoft

    Open source projects and samples from Microsoft.

  • Google photo Google

    Google ❤️ Open Source for everyone.

  • D3 photo D3

    Data-Driven Documents codes.