rrupam Goto Github PK

2017_08_08_test

aws-apps-scripts

An interface to authenticate AWS api requests from within google scripts

backtranslated-imdb

Backtranslations of IMDB movie reviews for Data Augmentation Purposes

bert_sentence_classification

Implement a plain vanilla BERT for multiclass classification. Can be used for reviews/comments/intent classification

best-readme-template

An awesome README template to jumpstart your projects!

bias-in-toxicity-classification

We write an algorithm that detects and removes unintended bias in that occurs in toxic comment classification. For instance, when we train for classification of toxic comments, the network may learn to associate toxicity with certain non-toxic features. For instance, because the word 'gay' is used frequently in a toxic comment and less in a positive or neutral comment, the network may mistakenly learn to associate the word 'gay' with toxicity so that when it encounters a neutral statement like 'I am a gay man', it may mistakenly identify the comment as toxic. This algorithm seeks to address this problem. The challenge is a Kaggle competition - https://www.kaggle.com/c/jigsaw-unintended-bias-in-toxicity-classification.

carlos-aguayo.github.io

deploying_sentiment_analysis

Building, training and deploying using Amazon's AWS SageMaker

detecting_skin_cancer

dog-breed-classification-project

We create a neural network to classify the breed of a given dog, using its image. We use two approaches - one is a CNN built from scratch. The second is a CNN using transfer learning. As an added bonus, we take human faces and categorize which breed it most closely resembles. ;)

generate_faces

We use Generative Adversarial Networks (GANs) to generate faces, given the data CelebA, a dataset consisting of around 200,000 images of celebrity faces.

generate_tv_scripts

We use LSTMs to generate TV scripts, given a starting word. As training data, we used Seinfeld's scripts.

graphql-to-gsheet

GraphQL requests beautifully presented in a Google Sheet

hello-world

Repository

nlp_tutorials

pandas

Reference material for basic and advanced Pandas operations in Python

policy-toolkit

Massively multitask fine tuning of roBERTa for policy priority identification in unstructured text

pretrainedcnn_clustering

ptsd

python_codes

Basic python syntax

rasa-fastai

Rasa bot using fast.ai text for NLU intent classifier

regularization

We explore regularization techniques like Ridge, Lasso and Elastic net regularization.

sentiment_classification

We predict the sentiment of movie reviews, from scratch.

spotify_eda

We identify commonalities and correlations in Spotify's top 100 tracks

titanic-with-pytorch

Using Pytorch to solve the famous titanic dataset problem, or in other words, killing a fly with a tank.

wri