Toots for dbreunig@note.computer account

Written by Drew Breunig on 2025-01-22 at 23:59

A new term I’m hearing more often, “authentic data,” illustrates a concern among data buyers and users that machine-generated errors will lead them astray. https://www.dbreunig.com/2025/01/20/on-authentic-data.html

=> More informations about this toot | View the thread

Written by Drew Breunig on 2025-01-09 at 19:04

Making your own evals is essential if you're building with LLMs.

Here's a quick example eval and a demo of its utility when choosing models, optimizing prompts, and more: https://www.dbreunig.com/2025/01/08/evaluating-llms-as-knowledge-banks.html

=> More informations about this toot | View the thread

Written by Drew Breunig on 2025-01-01 at 21:20

Summed up best by @annaleen https://www.newscientist.com/article/mg26234952-100-what-everyone-gets-wrong-about-the-2015-ashley-madison-scandal/

=> More informations about this toot | View the thread

Written by Drew Breunig on 2025-01-01 at 21:13

I tend to think the Ashley Madison leak, where it was revealed most of the “women” on it were bots is the most foreshadowing story of this era.

Networks will continue to hack our perception like orchids tricking pollinators, so long as PM’s keeping using engagement as their key metric.

https://nymag.com/intelligencer/article/meta-wants-more-ai-bots-on-facebook-and-instagram.html

=> More informations about this toot | View the thread

Written by Drew Breunig on 2024-12-29 at 21:19

OpenAI is Netscape is a really interesting take. https://mastodon.social/@daringfireball/113738152855628766

=> More informations about this toot | View the thread

Written by Drew Breunig on 2024-12-12 at 22:41

Really enjoy DSPy’s workflow for LLM work. Handing off the specifics of prompt generation and engineering back to the LLM makes a lot of sense: https://www.dbreunig.com/2024/12/12/pipelines-prompt-optimization-with-dspy.html

=> More informations about this toot | View the thread

Written by Drew Breunig on 2024-12-06 at 20:50

I want a speaker that:

• Isn't portable (no battery)

• Supports AirPlay

• Has no microphones

It is surprisingly difficult to find!

=> More informations about this toot | View the thread

Written by Drew Breunig on 2024-12-05 at 22:28

Reviewing the history of machine learning, we can both understand how LLMs advanced so quickly and why they've hit a wall.

We used 3 decades of Internet data and graphics innovation all at once. It's slow going from here on out.

https://www.dbreunig.com/2024/12/05/why-llms-are-hitting-a-wall.html

=> More informations about this toot | View the thread

Written by Drew Breunig on 2024-12-05 at 03:03

When You're Raising: "It is possible that we will have superintelligence in a few thousand days."

When You Want Out of a Profit-Sharing Contract w/ Microsoft: "My guess is we will hit AGI sooner than most people in the world think and it matter much less."

https://www.theverge.com/2024/12/4/24313130/sam-altman-openai-agi-lower-the-bar

=> More informations about this toot | View the thread

Written by Drew Breunig on 2024-11-21 at 16:41

What a novel idea. https://www.theverge.com/2024/11/21/24302400/threads-algorithm-focus-people-you-follow

=> More informations about this toot | View the thread

Written by Drew Breunig on 2024-10-29 at 19:01

It's a static site that's updated hourly with Github Actions. You can tweak the code to your heart's content. Here's the write up: https://www.dbreunig.com/2024/10/29/generating-descriptive-weather-forecasts-with-llms.html

=> More informations about this toot | View the thread

Written by Drew Breunig on 2024-10-29 at 19:01

Here's a little LLM demo using a model to generate weather reports based off webcam images and current conditions: https://sfweather.dbreunig.com

=> More informations about this toot | View the thread

Written by Drew Breunig on 2024-10-29 at 01:57

“My ‘I am Principled’ op-ed has a lot of people asking questions already answered by my op-ed.”

=> More informations about this toot | View the thread

Written by Drew Breunig on 2024-10-18 at 20:49

Simplifying the AI noise by segmenting everything into 3 big use cases: Gods, Interns, and Cogs. https://www.dbreunig.com/2024/10/18/the-3-ai-use-cases-gods-interns-and-cogs.html

=> More informations about this toot | View the thread

Written by Drew Breunig on 2024-10-14 at 18:33

Here's mine by the way, made probably 25 laters:

=> View attached media

=> More informations about this toot | View the thread

Written by Drew Breunig on 2024-10-14 at 18:16

Bonham's latest History of Science & Technology auction is up. The highlight for me is this Blue Box from 1972: https://www.bonhams.com/auction/29514/lot/31/blue-box-phone-hacking-device-blue-box-c-1972/

=> View attached media | View attached media

=> More informations about this toot | View the thread

Written by Drew Breunig on 2024-09-30 at 17:10

$160 billion dollars invested.

=> View attached media | View attached media | View attached media | View attached media

=> More informations about this toot | View the thread

Written by Drew Breunig on 2024-09-30 at 17:00

Joining Overture Places with county health restaurant grades using DuckDB, Ollama, and more. https://www.dbreunig.com/2024/09/27/conflating-overture-points-of-interests-with-duckdb-ollama-and-more.html

=> More informations about this toot | View the thread

Written by Drew Breunig on 2024-09-23 at 19:09

What would an AI collapse leave behind? https://www.dbreunig.com/2024/09/23/what-remains-if-the-ai-bubble-bursts.html

=> More informations about this toot | View the thread

Written by Drew Breunig on 2024-09-20 at 19:43

So…………what iOS podcast player do you all recommend?

=> More informations about this toot | View the thread

=> This profile with reblog | Go to dbreunig@note.computer account