Dave Craige
RT @TheStudio_io: Keep growing 🌱
tweet

anton
By far easiest way to reproduce model inference env is docker, I want local to mimic my deployment on runpod

docker run --gpus all -it -p 8888:8888 runpod/pytorch:3.9-1.13.1-116 /bin/bash

1. clone code to container
2. vscode attach to container
3. write code/run model
4. commit
tweet

Travis Fischer
RT @transitive_bs: Old ⇒ new https://t.co/JHeS0ZAXe3
tweet

Stephen Young
RT @alexqgb: At least half my retweets from the last day or so are showing "tweet unavailable". Clicking through anyway reveals that the tweet is, in fact, available. So yeah, maybe firing half the staff wasn't such a hot idea.

Meanwhile, the chill vibes on Mastodon continue to grow.
tweet

Stable Diffusion 🎨 AI Art
Public Accounting as a renaissance painting #AIArt

#StableDiffusion2 / #StableDiffusion #DreamStudio https://t.co/ELVjiqJ13z
tweet

CoffeeVectors
RT @bleedingedgeai: Microsoft is releasing the new Bing search as a chat interface (powered by gpt4).

Its rumored to be able to:
- search and cite sources
- perform complex tasks like building a travel plan
- be creative
- ask you follow up questions https://t.co/vEgNKHhu6P
tweet

Earth Is A Sales Funnel For SATAN
RT @goth600: Long been hailed as “The Most Important Century”; if there was ever a time in your entire genetic ancestry to kick it into high agency and create powerful relations with your local optima of players & entities, it’s now.
tweet

Robert Scoble
My AI driven car drove me to an AI party of AI researchers from Stanford while I was listening to AI artists on the way to eat AI-designed cakes while talking with AI entrepreneurs and using AI to make this neural radiance field.

https://t.co/ilJ4BuPKCI
tweet

Daniel Vassallo
I'm writing code again https://t.co/ULK4LUwidH
tweet

Jeremy Torman (editions available now!)
Another Roadside Attraction https://t.co/EUWrS8yCuH
tweet

Stephen Young
RT @weavinglightart: 🐲☕✨ https://t.co/vb2de75h3Y
tweet

murat 🍥
fine-tuning language models will go away

it will be replaced with running many consecutive prompts to build up a context embedding that can generate your desired outputs on a foundation model directly

and i think it will probably come this year
tweet