An Ahead of AI article by Sebastian Raschka sharing lessons from hundreds of experiments on finetuning LLMs with low-rank adaptation (LoRA), billed as an essential technique for efficiently training custom open-source LLMs. It covers LoRA consistency, QLoRA compute-memory trade-offs, learning rate schedulers, Adam versus SGD, multiple epochs, enabling LoRA for more layers, and balancing rank and alpha.
TextRazor is a natural language processing API that extracts meaning from text. It performs entity extraction, disambiguation and linking, keyphrase extraction, and automatic topic tagging and classification across 19 languages. It classifies content to the IAB Content and IPTC Media Topic taxonomies and analyzes relations and typed dependencies. Official client SDKs exist in Python, Java and PHP, with a REST API for other languages.
A service for generating AI profile pictures that is now available in beta. Users purchase credit to generate over 200 images, upload at least 10 photos, and wait 2-3 hours for generation.
Magick is described as a visual IDE for programming and deploying AI agents, using a node-based interface. It is platform agnostic and open source, and is noted as being in beta. It includes document processing and real-time, event-driven behavior.
Kaveen Kumarasinghe is a software engineer in Seattle working at Stripe and previously part-time at OpenAI, building open-source tools and AI experiments. The site lists projects including GPTDiscord, an open-source Python integration that brings GPT conversations, image generation and document questions into Discord.
Relevance AI is a low/no-code platform for building AI agents and multi-agent teams that autonomously complete tasks. Agents can be built with a no-code builder, pre-built templates or by cloning from a marketplace. The platform also includes a chat interface, a RAG-based knowledge feature, a no-code workflow builder for integrations and a marketplace.
astroML is a Python module for machine learning and data mining built on numpy, scipy, scikit-learn, matplotlib and astropy, distributed under the 3-clause BSD license. It provides statistical and machine learning routines for analyzing astronomical data, loaders for open astronomical datasets, and examples of analyzing and visualizing datasets. The project was started in 2012 to accompany a textbook on statistics, data mining and machine learning in astronomy.
Community notes
Be the first to share how this project worked for you.
Sign in with a verified account to leave a review or join the discussion.
Sign in to reviewKeep exploring
TextRazor is a natural language processing API that extracts meaning from text. It performs entity extraction, disambiguation and linking, keyphrase extraction, and automatic topic tagging and classification across 19 languages. It classifies content to the IAB Content and IPTC Media Topic taxonomies and analyzes relations and typed dependencies. Official client SDKs exist in Python, Java and PHP, with a REST API for other languages.
A service for generating AI profile pictures that is now available in beta. Users purchase credit to generate over 200 images, upload at least 10 photos, and wait 2-3 hours for generation.
Magick is described as a visual IDE for programming and deploying AI agents, using a node-based interface. It is platform agnostic and open source, and is noted as being in beta. It includes document processing and real-time, event-driven behavior.
Kaveen Kumarasinghe is a software engineer in Seattle working at Stripe and previously part-time at OpenAI, building open-source tools and AI experiments. The site lists projects including GPTDiscord, an open-source Python integration that brings GPT conversations, image generation and document questions into Discord.
Relevance AI is a low/no-code platform for building AI agents and multi-agent teams that autonomously complete tasks. Agents can be built with a no-code builder, pre-built templates or by cloning from a marketplace. The platform also includes a chat interface, a RAG-based knowledge feature, a no-code workflow builder for integrations and a marketplace.
astroML is a Python module for machine learning and data mining built on numpy, scipy, scikit-learn, matplotlib and astropy, distributed under the 3-clause BSD license. It provides statistical and machine learning routines for analyzing astronomical data, loaders for open astronomical datasets, and examples of analyzing and visualizing datasets. The project was started in 2012 to accompany a textbook on statistics, data mining and machine learning in astronomy.
