Sign in

Nathan Cooper

@ncoop57.bsky.social
465 followers 2.6K following 46 posts

I'm a nerd (nathancooper.io). The world can be ugly and cruel to the most innocent. Consider donating to help children suffering from one of the worst things: thorn.org/donate

PostsRepliesMedia
Nathan Cooper @ncoop57.bsky.social · 23/07/2025
At @answerdotai, we integrate @stripe into lots of projects. Every time, I found myself doing the same dance: create product, create price, create checkout session. Then hunting docs for parameters for each. So we built FastStripe, a self-documenting Stripe SDK that's easy to use!
A code snippet in Python demonstrating the use of the faststripe library to create a Stripe payment. The code imports StripeApi, initializes it with a placeholder key, and then calls the one_time_payment method to create a checkout session for a 'Digital Course' costing 4999 cents, with specified success and cancel URLs. The final line prints the first 64 characters of the generated checkout URL.
120
Nathan Cooper @ncoop57.bsky.social · 20/03/2025
🚀 ShellSage 0.1.0 has arrived, much to the relief of shell-challenged humans everywhere! Introducing sagent mode: now you can manipulate files by simply asking instead of learning actual commands. pip install shell-sage and embrace your laziness with ssage --mode agent today!
020
Nathan Cooper @ncoop57.bsky.social · 22/02/2025
🎉 ShellSage 0.0.8 is now live! New features: • Query logging to SQLite using `--log` flag (because who doesn't love a good paper trail?) 🗃️ • Better tmux history handling (for when your terminal has an existential crisis) 🛠️ Get it now: pip install shell_sage
120
Reposted by Nathan Cooper
Benjamin Warner @benjaminwarner.dev · 10/02/2025
With @bclavie.bsky.social and @ncoop57.bsky.social, we tried to answer two questions: - Can an instruction-tuned ModernBERT zero-shot tasks using the MLM-head? - Could we then fine-tune instruction-tuned ModernBERT to complete any task? Detailed answers: arxiv.org/abs/2502.03793
arxiv.org
It's All in The [MASK]: Simple Instruction-Tuning Enables BERT-like Masked Language Models As Generative Classifiers
While encoder-only models such as BERT and ModernBERT are ubiquitous in real-world NLP applications, their conventional reliance on task-specific classification heads can limit their applicability com...
141
Nathan Cooper @ncoop57.bsky.social · 01/02/2025
🤖 ShellSage v0.0.7 What's new? (besides our growing disdain for user error) • Added --v flag because apparently you need to know these things • Fixed config issues because reading docs is hard • Better tmux support (you're welcome) Upgrade your terminal's IQ: `pip install shell-sage`
140
Nathan Cooper @ncoop57.bsky.social · 01/02/2025
Had another magical moment with shell sage! I was fixing a repo issue and asked shell sage for the issues. Then used my quick commit alias (which uses shell sage), and it automatically added “Resolves #22” to the commit since it could see them in my terminal! 🤓🤓
120
Nathan Cooper @ncoop57.bsky.social · 19/01/2025
Found an amazing new usecase for shellsage's command mode: autogenerating git commits 🤓! I aliased `qc` to `git diff --cached | ssage --c please generate a concise git commit for these changes` and shellsage will enter the command I need with a great commit msg!
130
Reposted by Nathan Cooper
Nathan Cooper @ncoop57.bsky.social · 28/12/2024
🎉 ShellSage v0.0.6 is now available! New features: • Add new command mode for automatically inserting the correct command into your terminal • Add Support for OpenAI Compatible Providers Various bugs squashed too! github.com/AnswerDotAI/...
0103
Nathan Cooper @ncoop57.bsky.social · 28/12/2024
🎉 ShellSage v0.0.6 is now available! New features: • Add new command mode for automatically inserting the correct command into your terminal • Add Support for OpenAI Compatible Providers Various bugs squashed too! github.com/AnswerDotAI/...
0103
Reposted by Nathan Cooper
Benjamin Warner @benjaminwarner.dev · 22/12/2024
Thanks to my two co-leads: @nohtow.bsky.social , @bclavie.bsky.social , & the rest of our stacked author cast: @orionweller.bsky.social, Oskar Hallström, Said Taghadouini, Alexis Gallagher, Raja Biswas, Faisal Ladhak, @tomaarsen.com , @ncoop57.bsky.social , Griffin Adams, @howard.fm , & Iacopo Poli
161
Nathan Cooper @ncoop57.bsky.social · 21/12/2024
Interesting use of Shell Sage to see all of the panes that helped me out today. Had two panes: one for setting up Caddy configuration, and one for running a FastHTML app. Was getting a 303 error that I didn't understand, and so asked Shell Sage what was happening and gave it the entire context🤓
050
Reposted by Nathan Cooper
Nathan Cooper @ncoop57.bsky.social · 19/12/2024
🎉 ShellSage v0.0.5 is now available! New features: • ShellSage defaults to seeing entire tmux scrollback history • Important settings configurable via ~/.config/shell_sage/shell_sage.conf • System info now included in prompts pip install shell-sage github.com/AnswerDotAI/...
github.com
GitHub - AnswerDotAI/shell_sage: ShellSage saves sysadmins’ sanity by solving shell script snafus super swiftly
ShellSage saves sysadmins’ sanity by solving shell script snafus super swiftly - AnswerDotAI/shell_sage
141
Reposted by Nathan Cooper
Jeremy Howard @howard.fm · 19/12/2024
Post credits easter egg: Hey did you wonder what if we trained a bigger model? Where would that take us? Yeah, us too. So we're gonna train a "huge" version of this model in 2025. We might need to change the y-axis on this graph…
4174
Reposted by Nathan Cooper
Jeremy Howard @howard.fm · 19/12/2024
I'll get straight to the point. We trained 2 new models. Like BERT, but modern. ModernBERT. Not some hypey GenAI thing, but a proper workhorse model, for retrieval, classification, etc. Real practical stuff. It's much faster, more accurate, longer context, and more useful. 🧵
19620147
Reposted by Nathan Cooper
Benjamin Warner @benjaminwarner.dev · 13/12/2024
I feel the need for speed.
381
Nathan Cooper @ncoop57.bsky.social · 19/12/2024
🎉 ShellSage v0.0.5 is now available! New features: • ShellSage defaults to seeing entire tmux scrollback history • Important settings configurable via ~/.config/shell_sage/shell_sage.conf • System info now included in prompts pip install shell-sage github.com/AnswerDotAI/...
github.com
GitHub - AnswerDotAI/shell_sage: ShellSage saves sysadmins’ sanity by solving shell script snafus super swiftly
ShellSage saves sysadmins’ sanity by solving shell script snafus super swiftly - AnswerDotAI/shell_sage
141
Reposted by Nathan Cooper
Simon Willison @simonwillison.net · 09/12/2024
I can now run a GPT-4 class model on my laptop (The exact same laptop that could just about run a GPT-3 class model 20 months ago) The new Llama 3.3 70B is a striking example of the huge efficiency gains we've seen in the last two years simonwillison.net/2024/Dec/9/l...
simonwillison.net
I can now run a GPT-4 class model on my laptop
Meta’s new Llama 3.3 70B is a genuinely GPT-4 class Large Language Model that runs on my laptop. Just 20 months ago I was amazed to see something that felt …
1135859
Reposted by Nathan Cooper
Isaac Flath @isaac-flath.bsky.social · 07/12/2024
The code is quite simple! Check it out at gallery.fastht.ml/split/widget...
gallery.fastht.ml
FastHTML Gallery
A gallery of FastHTML components showing common patterns in FastHTML apps, including chat bubbles, cascading dropdowns, interactive charts, etc.
021
Reposted by Nathan Cooper
Isaac Flath @isaac-flath.bsky.social · 06/12/2024
🎉 New FastHTML Gallery entry thanks to @carlolepelaars.bsky.social HTMX example demonstrating click to load new table row 😃
184
Reposted by Nathan Cooper
Isaac Flath @isaac-flath.bsky.social · 07/12/2024
Thanks to HTMX + fh-frankenui the implementation is very simple! Check it out :D gallery.fastht.ml/split/dynami...
gallery.fastht.ml
FastHTML Gallery
A gallery of FastHTML components showing common patterns in FastHTML apps, including chat bubbles, cascading dropdowns, interactive charts, etc.
011
Reposted by Nathan Cooper
Isaac Flath @isaac-flath.bsky.social · 07/12/2024
🎉 New FastHTML Gallery entry! Dynamic dropdown with: ✅ Search box in dropdown ✅ Input for users to add additional values
121
Nathan Cooper @ncoop57.bsky.social · 07/12/2024
New version of shellsage (0.0.4)! now available on Pypi: - Add support for OpenAI models - Remove action mode - Add ability to configure shell sage with a configuration file pip install shell_sage ssage Hi --provider openai --model gpt-4o-mini
150
Reposted by Nathan Cooper
Nathan Cooper @ncoop57.bsky.social · 05/12/2024
As R&D staff @ answer.ai, I work a lot on boosting productivity with AI. A common theme that always comes up is the combination of human+AI. This combination proved to be powerful in our new project ShellSage, which is an AI terminal buddy that learns and teaches with you. A 🧵
77416
Reposted by Nathan Cooper
Omar Sanseviero @osanseviero.bsky.social · 06/12/2024
So how was my first week back at Google? - PaliGemma2 was openly released - DeepMind openly released GenCast weather forecast (in a GCP bucket 🙈) - Genie 2 🧞world model was announced This team ships 🔥🚀 Let's go!
1291
Reposted by Nathan Cooper
Isaac Flath @isaac-flath.bsky.social · 05/12/2024
ShellSage took me from swearing at AWS/GCP services to deciding to "just" set up my own server with a reverse proxy to host all my apps. I use ShellSage as a learning aid, and the boost in my learning speed in a huge unlock for me. A must have tool.
042
Reposted by Nathan Cooper
Jeremy Howard @howard.fm · 05/12/2024
I can't begin to describe how life-changing this new project, ShellSage, has been for me over the last few weeks. ShellSage is an LLM that lives in your terminal. It can see what directory you're in, what commands you've typed, what output you got, & your previous AI Q&A's.🧵
518521
Nathan Cooper @ncoop57.bsky.social · 05/12/2024
As R&D staff @ answer.ai, I work a lot on boosting productivity with AI. A common theme that always comes up is the combination of human+AI. This combination proved to be powerful in our new project ShellSage, which is an AI terminal buddy that learns and teaches with you. A 🧵
77416
Reposted by Nathan Cooper
Paige Bailey (webpaige.dev) @dynamicwebpaige.bsky.social · 04/12/2024
🎶 Google's Labs team has teamed up with Spotify on the first-ever 2024 Spotify 'Wrapped AI Podcast', built with NotebookLM! Hear AI hosts chat about your fave music, find out your listening trends and more → goo.gle/49lCpSC
1174
Reposted by Nathan Cooper
Simon Willison @simonwillison.net · 29/11/2024
This demo of structured data extraction running on an LLM that executes entirely in the browser (Chrome only for the moment since it uses WebGPU) is amazing My notes here: simonwillison.net/2024/Nov/29/...
simonwillison.net
Structured Generation w/ SmolLM2 running in browser & WebGPU
Extraordinary demo by Vaibhav Srivastav. Here's Hugging Face's [SmolLM2-1.7B-Instruct](https://huggingface.co/HuggingFaceTB/SmolLM2-1.7B-Instruct) running directly in a web browser (using WebGPU, so r...
518523
Reposted by Nathan Cooper
Dr Abeba Birhane @abeba.blacksky.app · 28/11/2024
Friends, this week marks a monumental step forward in my path towards pushing for a greater ecology of accountability in the age of AI. Thrilled to officially announce the AI Accountability Lab (AIAL) @theaial.bsky.social at Trinity College Dublin is launching today, Thursday Nov 28, 2024 aial.ie
aial.ie
AI Accountability Lab
Stewarding a greater ecology of accountability in the age of AI
911253310
Reposted by Nathan Cooper
Jeremy Howard @howard.fm · 28/11/2024
A librarian that previously worked at the British Library created a relatively small dataset of bsky posts, hundreds of times smaller than previous researchers, to help folks create toxicity filters and stuff. So people bullied him & posted death threats. He took it down. Nice one, folks.
2858258
Reposted by Nathan Cooper
Daniel van Strien @danielvanstrien.bsky.social · 26/11/2024
First dataset for the new @huggingface.bsky.social @bsky.app community organisation: one-million-bluesky-posts 🦋 📊 1M public posts from Bluesky's firehose API 🔍 Includes text, metadata, and language predictions 🔬 Perfect to experiment with using ML for Bluesky 🤗 huggingface.co/datasets/blu...
huggingface.co
bluesky-community/one-million-bluesky-posts · Datasets at Hugging Face
We’re on a journey to advance and democratize artificial intelligence through open source and open science.
69852772
Reposted by Nathan Cooper
Ethan Mollick @emollick.bsky.social · 26/11/2024
What if Clippy, but good? I asked Claude with computer use to watch me try to download an academic paper through the web and then give me advice on how to do it better & do it itself. Then I asked it to help me understand how to take good notes on a scientific paper & do that. Pretty sophisticated
2654
Reposted by Nathan Cooper
Jeremy Howard @howard.fm · 27/11/2024
I'm glad @hf.co is doing this. It brings down the barriers to allow more people to benefit from AI, rather than keeping it exclusively in the realm of deep pocketed giant companies. AI can help open the gates, to allow regular people to do things they couldn't do before. (Which can be threatening!)
916813
Reposted by Nathan Cooper
Jeremy Howard @howard.fm · 26/11/2024
People imagine that maintaining a Linux box requires dedicated sysadmins and lots of time. But actually if you’re doing normal stuff in a normal way, you can just leave it be, as long as you set it up right in the first place.
10843
Reposted by Nathan Cooper
merve @merve.bsky.social · 26/11/2024
Small yet mighty! 💫 We are releasing SmolVLM: a new 2B small vision language made for on-device use, fine-tunable on consumer GPU, immensely memory efficient 🤠 We release three checkpoints under Apache 2.0: SmolVLM-Instruct, SmolVLM-Synthetic and SmolVLM-Base huggingface.co/collections/...
1115927
Reposted by Nathan Cooper
Simon Willison @simonwillison.net · 25/11/2024
Anthropic released an interesting thing today: an attempt at a standard protocol for LLM tools to talk to services that provide tools and extra context to be used other the models modelcontextprotocol.io
modelcontextprotocol.io
Introduction - Model Context Protocol
Get started with the Model Context Protocol (MCP)
1837652
Reposted by Nathan Cooper
Ethan Mollick @emollick.bsky.social · 26/11/2024
A blindspot for AI reasoning engines like o1 is that they all appear to be trained on very traditional deductive problem solving for chain of thought What would a model trained on induction or abduction do? What about one trained on free association? Expert heuristics? Randomized exquisite corpse?
61178
Reposted by Nathan Cooper
Charlie Marsh @crmarsh.com · 19/11/2024
The latest uv release includes support for conflicting dependencies across optional groups. A subtle but very powerful feature. For example: use the PyTorch CPU build with `uv sync --extra cpu` and the CUDA build with `uv sync --extra gpu`. All powered by a single lockfile.
[project]
name = "project"
version = "0.1.0"
requires-python = ">=3.10.0"

[project.optional-dependencies]
cpu = ["torch==2.5.1+cpu"]
gpu = ["torch==2.5.1+cu124"]

[tool.uv]
conflicts = [
  [
    { extra = "cpu" },
    { extra = "gpu" },
  ],
]
29210
Reposted by Nathan Cooper
Charlie Marsh @crmarsh.com · 19/11/2024
uv now ships with dedicated documentation for PyTorch
"Using uv with PyTorch"
414713