Stay Confident
Subscribe to our weekly newsletter to stay confident in the AI systems you build.
An Introduction to LLM Benchmarking
LLM benchmarks like MMLU and HellaSwag score models on standardized tasks. Learn what they measure, where they fall short, and how to choose the right LLM.
Jeffrey Ip
A Step-By-Step Guide to Evaluating an LLM Text Summarization Task
Evaluate LLM text summarization by building a custom metric that scores factual alignment and coverage against the source, with working DeepEval code examples.
Jeffrey Ip
Why OpenAI Assistants is a Big Win for LLM Evaluation
We built JudgmentalGPT, an in-house LLM-as-a-judge evaluator, on OpenAI's Assistants API. Here's why Assistants is a big win for LLM evaluation, with code.
Jeffrey Ip
Become a Prompt Artist: Understanding the Midjourney LLM
Midjourney turns text prompts into images. Learn how it interprets prompts and the structure, parameters, and phrasing that give you precise control over the output.
Jeffrey Ip
How to Evaluate LLM Applications: The Complete Guide
Learn how to evaluate LLM applications and RAG pipelines the right way: the metrics that matter, common pitfalls, and a workflow you can run with DeepEval.
Jeffrey Ip
Why we replaced Pinecone with PGVector
We replaced Pinecone with pgvector for our GenAI app. Do you actually need a dedicated vector database? Our take on the cost, performance, and simplicity tradeoffs.
Jeffrey Ip
What is Retrieval Augmented Generation (RAG)?
Retrieval-augmented generation (RAG) grounds LLM answers in external data by pairing a retriever that fetches relevant context with a generator that writes the response.
Jeffrey Ip
A Gentle Introduction to LLM Evaluation
A gentle introduction to LLM evaluation: what it is, why it matters, and the core metrics and methods for scoring LLM outputs automatically at scale.
Jeffrey Ip
How to build a PDF QA chatbot using OpenAI and ChromaDB
Learn how to build a RAG-based Q&A chatbot over your own PDFs using OpenAI and ChromaDB, with step-by-step code for chunking, embeddings, and retrieval.
Jeffrey Ip
Building a customer support chatbot using GPT-3.5 and lLamaIndex
Build a customer support chatbot with GPT-3.5 and LlamaIndex: ingest your docs, add retrieval over them, and return grounded answers from your own knowledge base.