JavaScript is not available.

Sep 6, 2022

I’ve been waiting for over a year to share this.. Today we're launching Steamship, the fastest way to add language AI to a product.

Quote Tweet

Steamship

Sep 6, 2022

Hello, World! We’re Steamship — the fastest way to add language AI to a product. We’re dedicating ourselves to putting language AI into the toolbox of every software developer, everywhere. Here’s why

Show this thread

Feb 4

Today some folks want a disclaimer that a response was AI generated. I think within two years some folks will a guarantee that human responses were properly “AI augmented”. The question is what do we call pure human output? Vegan or organic?

Replit is great for quick show-and-tell. Race condition I've got to iron out that sometimes strikes (hey, async ftw) But basically: runner = ship.use('async-prompt-chain-experiment') runner.invoke('create-chain', {..}) runner.invoke('run-chain', {..}) replit.com/@TedBenson/Clo

Quote Tweet

I see why prompt chaining is so addictive. Just added DALL-E support to this little async-chain experiment. Here's a minimal-effort chain that generates mysterious images from a topic like 'cake'

I see why prompt chaining is so addictive. Just added DALL-E support to this little async-chain experiment. Here's a minimal-effort chain that generates mysterious images from a topic like 'cake'

What's cool about this is that it isn't done synchronously -- each step is it's own async task in the cloud, and you can get callbacks as they complete or fail.

Did a bit of hacking on cloud-hosted, declarative prompt chaining tonight. In about ~200 lines of Python I whipped up a Steamship package that can host and run prompt chains at scale -- pretty neat. Here's a Replit you can use to try it . replit.com/@TedBenson/Clo

Gigachads today don't appreciate how expensive it was to be a Megachad back in the 90s.

Jan 31

Hey I've heard of some of these companies :)

Quote Tweet

Matt Shumer

@mattshumer_

Jan 31

Inference/Training/APIs: @BananaDev_ @gooseai_NLP @leap_api @GetSteamship @MosaicML @weights_biases

Show this thread

This is the first time I've seen a transformer model edit a document and regenerate it I am surprised more people aren't talking about this...

361

Excited about the possibilities for this and gaming bots! (And now I’m wondering what a natural language programming environment for in-game bots would look like…)

Quote Tweet

Regression Games

@RegressionGG

Jan 31

Day 30 of Gaming, AI, and Esports! There are so many fun ways to use AI and ML - one of these is NLP, especially with the new use cases being highlighted with GPT-3 and ChatGPT. You can now create your own NLP apps in minutes using @GetSteamship steamship.com #nlp

Show this thread

Refreshing Hacker News as you get ranked and reranked is funny because you find yourself competing neck-and-neck with Emacs, a touching Clown organization, and an astrophysics discovery.

Steamship is on the front page of Hacker News right now :) 🚢🚢🚢

Steamship is on the front page of Hacker News right now :) Check it out!

Thanks for the shoutout

@ai_kadhim

! We just launched it to Product Hunt today after iterating a ton with the early users the past two weeks! producthunt.com/posts/deploy-b

Quote Tweet

Ali Kadhim

@ai_kadhim

Jan 27

One-Click Deploy for GPT is here!!

The @GetSteamship team just launched the ability to 𝗱𝗲𝗽𝗹𝗼𝘆 𝗚𝗣𝗧 𝗔𝗣𝗜𝘀 𝘄𝗶𝘁𝗵 𝘁𝗵𝗲 𝗰𝗹𝗶𝗰𝗸 𝗼𝗳 𝗮 𝗯𝘂𝘁𝘁𝗼𝗻!

bit.ly/3kOAe52 #unautomatable #ai #chatgpt

Show this thread

Check us out on ProductHunt today! APIs built on this have already been trending for a week on Replit ❤️🚢

Quote Tweet

Steamship

Check us out on Product Hunt today! Build & deploy GPT-3 APIs in minutes. The fastest way to launch an LLM product. producthunt.com/posts/deploy-b

Jan 29

Holy moly. This is the first AI voice I would actually listen to alongside a human option. The fact that no corrections were made to the text is a BIG deal.

@elevenlabsio

do you have an API we can add support for on

Quote Tweet

AI narrates The Great Gatsby. Listen to a fragment from the classic by F. Scott Fitzgerald. Narrated by a fully AI-generated voice. No corrections were made.

Show this thread

2:14

50.9K views

Jan 29

This needs to be said more. LLMs aren’t the death of software engineering. They’re the platform that will 10x the reach of software engineering.

Quote Tweet

Scott Stevenson

@scottastevenson

Jan 28

“Are GPT3-based apps defensible?” is similar to asking “Are SQL database apps defensible?”

Jan 28

My son and I are pretending to start a nursery rhyme band called Stinktoven. Smash like and subscribe.

Jan 27

I have a soft spot for auto-generated UIs.. One of the fun little power-ups we added to

last week was auto-generating a dropdown box if your Prompt API uses an Enum as a parameter. Here's the full auto-generated demo: steamship.com/packages/insul

Jan 27

This is super cool. Also fascinating at the asymmetries in how embeddings of this size are both zero-cost and burdensome at the same time: - A few dollars to embed, but then - Over a gigabyte to store

Quote Tweet

Andrew Gao

@itsandrewgao

What if you could talk to the Bible? Try now: biblegpt.org Describe your situation or ask a question! Cites specific verses from the Bible. Works in English and Spanish. Made with @OpenAI and domain via @Porkbun! #biblegpt #chatgpt #gpt3

Show this thread

📦 One-Click Deploy for GPT is here!! 🛳 The

team just launched the ability to 𝗱𝗲𝗽𝗹𝗼𝘆 𝗚𝗣𝗧 𝗔𝗣𝗜𝘀 𝘄𝗶𝘁𝗵 𝘁𝗵𝗲 𝗰𝗹𝗶𝗰𝗸 𝗼𝗳 𝗮 𝗯𝘂𝘁𝘁𝗼𝗻! 🚀👇 bit.ly/3kOAe52 #unautomatable #ai #chatgpt

A Shazam bot built on

How to build a Shazam-like Telegram Bot using Python

! This is so cool

@AnelloEugenia

🚀 I didn't realize Telegram's API was so clean to interact with either..

towardsdatascience.com

A tutorial to create and deploy a telegram bot that takes in music in real-time, and helps you to search the title and the singer

The "recent inputs" feature on the web UIs each Prompt API generates is... pretty entertaining :)

This month has been RIDICULOUSLY fun --

@FraudulentTom

@eniascailliau

@jdouglasreid

, and I have been iterating on a way to build & deploy GPT/prompt APIs to the web. Just a few lines of python + `ship deploy` gets you: - Production-ready API - Web demo - Scaling across users

Quote Tweet

Steamship

If you’re into GPT, this is for you

The “Deploy” button for GPT APIs LLM backends are the future, we're building the way to launch & host them in minutes. If you’ve got a prompt, prompt-chain, or blend of Python & GPT, check it out! steamship.com/build/prompt-a

Jan 18

I'm not even going to click the link. It's too perfect just reading the title. I don't want to ruin my imagination of it.

Jan 17

New site for curating prompts. I've got a soft spot for Atari-style pixel art themes :)

Quote Tweet

Promptstacks

@Promptstacks

Jan 17

A #ChatGPT Prompt Engineering Community. Submit prompts and rank on a leaderboard. Product Hunt Release Day: producthunt.com/posts/promptst #ChatGPTGOD #chatgpt3 #chatgpt4

Question for Chicago boosters: What's stopping Chicago from reaching NYC levels of density and building a bunch more train lines?

221

GitHub - jeffbinder/promptarray: Text generator prompting with Boolean operators

Jan 10

I can't imagine some version of Jeff Binder's PromptArray NOT existing over time. Sure, LLM will need prompt hacking less & less.. But techniques that better instruct computers while enabling smaller models seem like a clear win even with GPT-10

github.com

Text generator prompting with Boolean operators. Contribute to jeffbinder/promptarray development by creating an account on GitHub.

Jan 9

VALL-E by Microsoft clones your voice with just a few seconds of audio. I wonder if LLMs + better voice gen gives new life to smart speaker apps, or if the whole format is just a weird dead end... valle-demo.github.io

Jan 9

This week The Diff by

@ByrneHobart

poses the fantastic question of whether AI is the new Visual Basic for Applications. Call me optimistic, but I think we'll be able to crunch LLMs down to the point where they fit on-device & this'll become a reality.

Jan 7

A while back I wrote a (true!) story about this here, from my grad school days in an AI lab: The Case of the Murderous AI edwardbenson.com/2018/09/case-o

Quote Tweet

Jan 7

It makes sense that AI will eat software because “software” is the the nature of its embodiment. You can only do so many RL loops with humans. You can do infinite RL loops with a REPL.

Jan 7

It makes sense that AI will eat software because “software” is the the nature of its embodiment. You can only do so many RL loops with humans. You can do infinite RL loops with a REPL.