Published:
Have a quick look into the Parameter-Efficient Fine-Tuning (PEFT) methods for Large Language Models (LLMs) and discover how these techniques are revolutionizing machine learning by optimizing computational resources without compromising performance. Read more
Published:
Have you encountered cases where inner product and dot product are used synonymously? If you have, be alert that inner product and dot product are not same. Read more
Published:
We’ll explore the synergy between AWS S3, Snowpark, and Snowflake for efficient handling of large tabular data. By combining these tools, you can seamlessly process and analyze extensive datasets stored in AWS S3. Read more
Published:
An end-to-end comprehensive scalable SageMaker pipeline for training, deploying and monitoring multiple models trained on different datasets to a single endpoint. Read more
Published:
Welcome to my latest blog post! Today, I am excited to share my recent exploration into the fascinating world of reinforcement learning, specifically focusing on the multi-armed bandit problem and its various solutions. As a foundation for my implementation, I closely followed the insightful book, Reinforcement Learning: An Introduction (second edition) by Richard S. Sutton and Andrew G. Barto. Read more
Published:
It is an annotated criticism on the paper Why do tree-based models still outperform deep learning on typical tabular data?. A personal view on why it was accepted for publication at NEURIPS 2022 is discussed along with a brief description of the key contributions and the significance of the findings presented in the paper. Also, some of the key strengths and weaknesses with this paper, the approach taken by the authors, and the clarity and ease of understanding of the writing are discussed herewith. Read more
Published:
Wav2Vec2 is a pretrained model for Automatic Speech Recognition (ASR) and was released in September 2020 by Alexei Baevski, Michael Auli, and Alex Conneau. Soon after the superior performance of Wav2Vec2 was demonstrated on one of the most popular English datasets for ASR, called LibriSpeech, Facebook AI presented a multi-lingual version of Wav2Vec2, called XLSR. XLSR stands for cross-lingual speech representations and refers to model’s ability to learn speech representations that are useful across multiple languages. Read more
Published:
Humans can recognize objects from a few examples. Having seen a lot of animal images before, given very few images of novel animals, we can recognize them easily. But for deep learning models, we have to train them from scratch to learn a new task. Read more