🔭 LangChain's State of AI 2023 report

📖 TUTORIALS & CASE STUDIES

Mastering Prompt Engineering for AI Applications

read time: 15 minutes
This tutorial provides a deep dive into prompt engineering for AI applications, using the example of an AI-based Crossword Puzzle Helper. It explains how to structure prompts for generative AI models, and demonstrates how to build a sentiment analysis bot using TypeScript and the LLaMa2 model.

Building a ChatGPT Clone with Google's Gemini API

read time: 20 minutes
This tutorial provides a comprehensive guide on using Google's Gemini API to build a ChatGPT-like application. It covers the process of obtaining a free Google API Key, installing necessary dependencies, and crafting code to build intelligent chatbots that can handle text, images, audio, and video.

Llama 2: Fine-tuning, Inference Recipes, and Demo Apps

read time: 15 minutes
The Llama 2 repository provides fine-tuning and inference recipes, demo apps, and support for Llama Guard safety checker. It also includes instructions for running Llama 2 locally, in the cloud, or on-prem, and integrating it with various platforms. Learn more about it here.

Master Advanced Retrieval Techniques for AI with Chroma

read time: 5 minutes
DeepLearning.AI offers a short course on advanced retrieval techniques for AI. The course covers query expansion, cross-encoder reranking, and embedding adapters to improve the relevancy of retrieved results. Ideal for those with intermediate Python knowledge looking to enhance their data retrieval skills. Free access for a limited time.

🧰 TOOLS

Top 9 AI APIs for Developers in 2024

read time: 7 minutes
This blog post curates the top AI APIs for 2024, including OpenAI, IBM Watson, Google AI, Microsoft Azure Cognitive Services, Amazon AI, Komprehend, MonkeyLearn, Clarifai, and Wit.ai. These APIs offer a range of services from natural language processing to image and video analysis, aiding developers in building smarter, more intuitive applications.

PostgresML: MLOps Platform in a PostgreSQL Extension

read time: 7 minutes
PostgresML is an MLOps platform built as a PostgreSQL extension, enabling developers to build AI applications directly in their database. It supports various use cases like chatbots, site search, fraud detection, and forecasting. The platform offers high efficiency, in-database ML, open-source libraries, and instant scalability. Explore more about PostgresML here.

UForm: A Compact Multimodal AI Library for Content Understanding and Generation

read time: 15 minutes
UForm is a versatile and efficient multimodal AI library that offers tiny embedding models for understanding and searching visual and textual content across various languages. It also provides small generative models for tasks like image captioning and Visual Question Answering (VQA). The library's compact pre-trained transformer models can run on various platforms, from servers to smartphones. Read more

Introducing diffusers.js: A Library for Running Diffusion Models on GPU/WebGPU

read time: 3 minutes
The diffusers.js library allows developers to run diffusion models on GPU/WebGPU. It includes a pre-trained model optimized for GPU, with examples provided for both browser and Node.js environments. The library requires CUDA/DML/WebGPU support, but also offers a 'cpu' revision for machines without GPU.

📰 NEWS & EDITORIALS

LangChain: State of AI 2023

read time: 15 minutes
LangChain provides a detailed analysis of Generative AI trends in 2023, highlighting the most used LLM providers, vectorstores, and embeddings. The report also discusses the rise of LangSmith, a cloud platform for prototyping and production, and the challenges in testing LLM applications. Read the full report here.

2023: The Breakthrough Year for Large Language Models

read time: 15 minutes
The year 2023 marked significant advancements in Large Language Models (LLMs), with their ease of construction and application in various fields, including coding, becoming increasingly apparent. However, challenges such as gullibility and ethical complexities persist. Read more about the year's AI developments in this comprehensive roundup.

Japan's Bold Move: No Copyright Enforcement on AI Training Data

read time: 5 minutes
In a significant policy shift, Japan has decided not to enforce copyrights on data used for AI training. This move is part of Japan's ambitious plan to become a global leader in AI technology. However, this decision has sparked concerns among artists who fear that AI could devalue their work. Read more about this development on BIIA.