So AI Won’t Scale, Now What?Rumours that the famed Transformer architecture that powers GPT, Claude and Llama, won’t keep getting better with more compute (the scaling…Nov 12Nov 12
Simplified Handwriting OCR with AnnoMemo Telegram BotEarlier this year I wrote about using VLM models to do OCR on my terribly scribbly hand writing. Models like GPT-4o are actually quite…Nov 3Nov 3
Running Phi MoE 3.5 on Macbook ProThe relatively recently released Phi 3.5 model series includes a mixture-of-experts model featuring 16 x 3.3 Billion parameter expert…Sep 5Sep 5
Ditch that Chatbot Subscription with Open Web UI — BrainsteamDitch that ChatGPT Subscription: Moving to Pay-as-you-Go AI usage with Open Web UIJul 8Jul 8
LLMs: To Fine-Tune or Not to Fine-Tune? — BrainsteamLLMs: To Fine-Tune or Not to Fine-Tune?May 7May 7
LLMs Can’t Do Probability — BrainsteamI built an experiment to show what happens when you ask an LLM to behave in a certain way a certain percentage of the time.May 1May 1
Self-hosting Llama 3 on a home serverSelf-hosting Llama 3 as your own ChatGPT replacement service using a 10 year old graphics card and open source components.Apr 208Apr 208
Reviewing the Best Paid and Open Source AI-Powered Handwriting OCRRemarkably (for me), this essay started its life as a scribble in a notebook rather than something I typed into a markdown editor!Apr 21Apr 21
How to use Airbyte Sync in ProductionSome tips for configuring airbyte for production-ready syncAug 15, 2023Aug 15, 2023
Published inFilament-SyfterChallenges and Opportunities for Generative AI for Private Market AnalystsThe recent wave of new generative AI solutions that leverage transformer-based large language models (LLMs) has injected renewed vigour and…Jun 13, 2023Jun 13, 2023