Share this page!

Enter URL or ID to Unroll

You can paste full URL like: https://x.com/threadreaderapp/status/1644127596119195649
or just the ID like: 1644127596119195649

How to get URL link on X (Twitter) App

  1. On the Twitter thread, click on or icon on the bottom
  2. Click again on or Share Via icon
  3. Click on Copy Link to Tweet
  4. Paste it above and click "Unroll Thread"!
  5. More info at Twitter Help
Travis Fischer Profile picture
Travis Fischer
@transitive_bs
Building a new type of @HackerResidency 💪 2 exits previously, @HF0, @Microsoft, @Amazon. My heart is open source 💕
2 subscribers
Mar 8 • 7 tweets • 3 min read
Roast your Twitter profile with AI-generated Severance wellness memes!

Some examples roasting @elonmusk 😂 Image
Image
Image
Image Create your own Severance memes based on your Twitter profile here:

Seriously, GPT-4.5's ability to generate hilarious, personal memes is uncanny...severance-wellness-session.com
Apr 3, 2023 • 11 tweets • 3 min read
A practical guide on how to use LLMs effectively: How to use LLMs effectively The key is to always start simple, and only add more complex approaches and tooling if necessary.

You'll be surprised at how far you can get with "prompt engineering" alone.

Apr 2, 2023 • 13 tweets • 4 min read
LLMs are stateless processors allowing for true decentralized compute at scale.

Combine that with external tools, and you have a fundamentally new computing paradigm.

This is the real web3 👇 HuggingGPT showing LLMs acting as controllers to external to This image of LLMs being used as controllers is from @huggingface's recent paper: arxiv.org/abs/2303.17580

It shows that LLMs work exceptionally well as controllers (or CPUs), displaying an emergent affinity to utilize external tools and models to accomplish complex tasks.
Apr 1, 2023 • 4 tweets • 4 min read
Projects I released in March 2023:

search.bensbites.co – Search the best resources in AI
chatgpthackers.dev – ChatGPT Hacker community
sklearn.vercel.app – ML library for Node.js
google-waitlist.vercel.app – Google Bard parody
apple-ai.vercel.app – Apple AI parody OSS projects I built in March 2023 All of these projects are open source, and they're all focused on AI.

In the spirit of building in public:

• my chatgpt npm repo passed 12k stars on github ⭐
• @ChatGPTBot passed 110k followers 🚀
• i'll likely have to shut down @ChatGPTBot due to twitter API changes 😞
Apr 1, 2023 • 4 tweets • 2 min read
For anyone frustrated with GPT-4's usage cap or slow speed in the ChatGPT webapp:

1. Sign up for gpt-4 api access: openai.com/waitlist/gpt-4…
2. Enter your api key into an OSS chatgpt clone github.com/mckaywrigley/c…
3. Enjoy GPT-4 that's uncapped, faster, and more powerful 🔥 The version of GPT-4 in the ChatGPT webapp is currently throttled and limited to 4k context length to help with ChatGPT's massive rollout.

The OpenAI Chat Completions API has two gpt-4 model variants that are faster and more powerful:

"gpt-4" model w/ 8k context and "gpt-4-32k"
Mar 15, 2023 • 16 tweets • 4 min read
Yesterday's GPT-4 launch was **crazy** 🔥

After discussing w/ colleagues and taking some time to collect my thoughts,

Here are the top 3 things GPT-4 enables that were previously impossible or extremely difficult to achieve – including examples 👇 1/ Image comprehension

⇒ Unlocks full-screen app and browser automation 🤖

We thought text-based co-pilot and writing assistants were cool, but now that GPT-4 natively understands images, automating app and browser workflows based on screenshots is going to skyrocket 🚀
Mar 13, 2023 • 10 tweets • 4 min read
Just released a new open source machine learning library for TypeScript 🔥

Node.js devs can now take full advantage of Python's powerful scikit-learn package – without needing to know any Python! 🤯

PCA, KMeans, TSNE, ... 257 classes in all!
→ github.com/transitive-bul… An open source TS package w... Here's a comparison of how to use the KMeans algorithm in Python vs TS to find clusters of related points in datasets.

The Python version on the left uses the official "scikit-learn" PyPI package, and the TypeScript version on the right uses the new "sklearn" NPM package. KMeans example
Feb 21, 2023 • 10 tweets • 6 min read
It's crazy how fast AI projects get stars on GitHub these days ⇒ OSS devs are very eager to play with quality AI projects.

Here are some of my favorite OSS AI projects from this past week: 🧵 All the goodness of @LangChainAI now in JS / TS!

github.com/cfortuner/prom… by @colinfortuner 🔥
github.com/hwchase17/lang… is a Node.js port of the popular Python project 🐍
Feb 20, 2023 • 5 tweets • 2 min read
Just pushed a major update to the chatgpt NPM package.

We now provide 3 ways of accessing ChatGPT: chatgpt NPM package exports 1. ChatGPTAPI uses text-davinci-003 with the official OpenAI completions API to mimic ChatGPT (we build prompts for you).

Pros:
• very customizable
• robust
• official
• lightweight

Cons:
* paid
• not using an LLM fine-tuned for chat
• quality isn't 100% on par w/ ChatGPT
Feb 9, 2023 • 12 tweets • 4 min read
I have access to Bing's new chat-based search 🔥

Will be posting interesting examples & learnings in this thread 🧵

Reply with your prompts / search queries and I'll take screenshots of Bing's answers. 1. Bing AI's self-described rules & limitations
Feb 5, 2023 • 10 tweets • 5 min read
I've been running @ChatGPTBot on twitter for the past two months, and it recently passed:

• 29k followers
• 100k public conversations 🔥
• 6.8k stars on github for the underlying NPM package

Here's a breakdown of the bot's usage and analytics over time: 🧵 ChatGPT Twitter Bot: Lessons from over 100k Conversations For starters, @ChatGPTBot is a simple twitter bot that waits for users to tag it with a question and then responds with an answer generated using ChatGPT's unofficial API.

Here's an example interaction:
Mar 5, 2021 • 8 tweets • 3 min read
Teachable co-founder @conradwa leading @ljin18's course today and dropping some serious knowledge bombs. 🔥

How to solve the cold-start problem via the Creator Growth Loop.

Thread 👇 1/ The Creator Growth Loop

Mar 4, 2021 • 6 tweets • 2 min read
I'm thoroughly enjoying @ljin18's creator economy course. 💯

It's forced me to really think about who these creators are from first principles.

Here are 3 ingredients that I think define successful creators: Independent, Sustainable, and Scalable. 1/ Successful creators are Independent.

They leverage their individuality to build a following and/or monetize their unique skills — without being tied to traditional institutions and full-time work.
Mar 2, 2021 • 5 tweets • 1 min read
Such a beautiful quote by famed statistician George Box. A quote by George Box As an analytical founder, I'm always trying to break down complex concepts into simplified models.

Lately, I've been experimenting with modeling the creator economy, different business models, etc.

Most of these models I end up throwing out, but a few end up being invaluable.
Feb 10, 2021 • 8 tweets • 2 min read
The creator economy is heating up! 🔥

What @josephpalbanese and his team are building over at @stir is truly next level.

So what are they doing well?

↓↓

theinformation.com/articles/andre… 1/ They've been able to secure key partnerships with influential thought leaders & creators from day one (tackling the market top down).

These people come with massive followings that Stir can leverage in so many different ways.
Feb 5, 2021 • 15 tweets • 3 min read
Taking notes on this clubhouse room with @allenlau on @wattpad recently acquired for $600M 🔥

#1 Clubhouse needs an incognito mode

Thread 👇 Allen wanted to "combine business with his passion" back in 2002 but was way too early for mobile tech at the time.

Joined his co-founder in 2006 when they realized they were working on the same thing independently.

=> convergent thinking makes for great co-founders