Dr James Ravenscroft – Medium

Dr James Ravenscroft

Published in

Managing Costs for Specialised Language Models

Oct 22

Diagram illustrating how VLLM efficiently hosts multiple LoRA adapters alongside a base language model. The diagram shows three user icons (labeled User A, User B, and User C) connecting to a central ‘VLLM Container’ which encapsulates a ‘Llama 3.3 70B’ base model. To the right of the base model are three blocks representing LoRA adapters: ‘Summary Adapter’, ‘Q&A Adapter’, and ‘Tag Generation Adapter’. Each adapter receives input from one or more of the user icons and outputs back to them. The

Oct 22

Voice Input Is Awesome

Introduction: The Promise and Disappointment of Early Voice Recognition

Apr 14

Voice Input Is Awesome

Apr 14

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 12, 2024

So AI Won’t Scale, Now What?

Nov 12, 2024

Simplified Handwriting OCR with AnnoMemo Telegram Bot

Earlier 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 3, 2024

Simplified Handwriting OCR with AnnoMemo Telegram Bot

Nov 3, 2024

Running Phi MoE 3.5 on Macbook Pro

The relatively recently released Phi 3.5 model series includes a mixture-of-experts model featuring 16 x 3.3 Billion parameter expert…

Sep 5, 2024

Running Phi MoE 3.5 on Macbook Pro

Sep 5, 2024

Ditch that Chatbot Subscription with Open Web UI — Brainsteam

Ditch that ChatGPT Subscription: Moving to Pay-as-you-Go AI usage with Open Web UI

Jul 8, 2024

Screenshot of a coding chat interface where the user requests a Python script with CLI options using the `click` module. The response provides a detailed guide, including installation instructions (`pip install click`) and a code snippet for processing an input file and saving the result to an output file. The left sidebar displays recent activity and categories, including today’s and previous days’ items like “Firms Embrace AI” and “Recipe Ingredients Transcription.” The footer contains a messa

Jul 8, 2024

LLMs: To Fine-Tune or Not to Fine-Tune? — Brainsteam

LLMs: To Fine-Tune or Not to Fine-Tune?

May 7, 2024

LLMs: To Fine-Tune or Not to Fine-Tune? — Brainsteam

May 7, 2024

LLMs Can’t Do Probability — Brainsteam

I 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 1, 2024

LLMs Can’t Do Probability — Brainsteam

May 1, 2024

Self-hosting Llama 3 on a home server

Self-hosting Llama 3 as your own ChatGPT replacement service using a 10 year old graphics card and open source components.

Apr 20, 2024

Self-hosting Llama 3 on a home server

Apr 20, 2024

Reviewing the Best Paid and Open Source AI-Powered Handwriting OCR

Remarkably (for me), this essay started its life as a scribble in a notebook rather than something I typed into a markdown editor!

Apr 2, 2024

Reviewing the Best Paid and Open Source AI-Powered Handwriting OCR

Apr 2, 2024

Dr James Ravenscroft

Dr James Ravenscroft

Ml and NLP Geek, Research Engineering Manager at Thomson Reuters. Saxophonist, foodie and explorer. I was born in Bermuda and I Live in the UK

Following

Help

Status

About

Careers

Press

Blog

Privacy

Rules

Terms

Text to speech