How to Use Integrated Gradients to Interpret BERT Model’s Prediction — Bidirectional Encoder Representation from Transformer or BERT is a language model that’s very popular within the NLP domain. BERT is literally the swiss army knife of NLP due to its versatility and how well it performed in many different NLP tasks, such as text classification, named entity recognition, question-answering, etc.

DVC: It’s a Git, but for Our Data and ML Model — I think we can all agree that every data science project has a high probability of becoming overwhelmingly messy. There are three common sources of mess in a data science project: code, data, and ML model. Here are some examples of the mess that originated either from data, ML model…

How to leverage a pre-trained BERT model for custom data to predict the entity of each word in a text — When it comes to dealing with NLP problems, BERT oftentimes comes up as a machine learning model that we can count on in terms of its performance. …

How to use LayoutParser library to detect the layout and extract texts from document images — The application of NLP oftentimes requires us to extract texts from input documents as prerequisites. The problem is, sometimes we need to do extra work to extract texts from the input documents because they normally come in PDF, JPEG, or PNG format. And this is where we usually use OCR…

How to find that perfect neural networks model for our use case with the help of Optuna — Developing the right neural network model can be time-consuming. As you might know, there are a lot of hyperparameters in a neural network model that we need to tweak to get that perfect fitting model such as the learning rate, optimizer, batch size, number of units in a layer, activation…

Build and deploy a simple machine translation web app with Spaces, Streamlit and Transformers — I think everybody agrees that being able to build meaningful projects is a prerequisite that we need to have in our portfolio in order to stand out when we want to break into a data science role. Fortunately, nowadays there are a lot of libraries, frameworks, and platforms that make…

How to leverage a pre-trained BERT model from Hugging Face to classify text of news articles — Back in 2018, Google developed a powerful Transformer-based machine learning model for NLP applications that outperforms previous language models in different benchmark datasets. And this model is called BERT. In this post, we’re going to use a pre-trained BERT model from Hugging Face for a text classification task. As you…

Should we trust our image classifier model? — The advancement rate and growth in the area of machine learning are insane. Nowadays, we can choose a variety of machine learning models to solve our problems. Let’s say we want to solve a classification task, now we don’t only have logistic regression to choose from. …

The concept of statistical inference, hypothesis testing, significance level, and p-value explained for dummies. — Learning statistics should be fun and intuitive, at least that’s what I think. However, when we try to learn statistics either in the graduate school or online, all of the technical jargon related to statistics are being thrown at us all of a sudden. Take a look at the definition…