This article presents and describes five commonly used prompt compression techniques to speed up LLM generation in challenging scenarios.
Author Archive | Iván Palomares Carrascosa
The Journey of a Token: What Really Happens Inside a Transformer
Let’s discover what happens inside a transformer model, that is, how input tokens or parts of an input text sequence turn into generated text outputs, and what is the rationale behind the changes and transformations that take place inside the transformer.
K-Means Cluster Evaluation with Silhouette Analysis
Learn how to evaluate K-means clustering quality using silhouette analysis with Python code examples.
Why Decision Trees Fail (and How to Fix Them)
Discover three common reasons why decision tree models fail and learn practical Python solutions to fix them.
Forecasting the Future with Tree-Based Models for Time Series
How to train decision tree models capable of performing forecasting on time series data by extracting lagged features and rolling averages.
Expert-Level Feature Engineering: Advanced Techniques for High-Stakes Models
Explore three expert-level feature engineering techniques for building robust, interpretable machine learning models in high-stakes applications.
The 7 Statistical Concepts You Need to Succeed as a Machine Learning Engineer
The seven core statistical pillars every machine learning engineer should master to build reliable intelligent systems.
How to Diagnose Why Your Language Model Fails
Learn a five-step framework to diagnose and fix common causes of language model failure.
10 Python One-Liners for Calculating Model Feature Importance
10 simple but effective Python one-liners to calculate model feature importance from different perspectives, enabling not only understanding of how your machine learning model behaves, but also why it predicts the way it does.
7 Prompt Engineering Tricks to Mitigate Hallucinations in LLMs
The 7 techniques listed in this article illustrate how both standalone LLMs and RAG system can improve their performance and become more robust against hallucinations by simply implementing them in your user queries.