I wanted to know how large the deepseek-ai/DeepSeek-V3-Base repo on Hugging Face was without actually downloading all of the files. With some help from Claude, here's the recipe that worked. First, clone the repo without having Git LFS download the files: GIT_LFS_SKIP_SMUDGE=1 git clone https://huggingface.co/deepseek-ai/DeepSeek-V3-Base cd...
I was exploring the original BERT model from 2018, which is mainly useful if you fine-tune a model on top of it for a specific task. dslim/distilbert-NER by David S. Lim is a popular implementation of this, with around 20,000 downloads from Hugging Face every month. I tried the demo from the README but it didn't quite work - it complained about...
I was getting the following warning for one of my Python test suites: DeprecationWarning: datetime.datetime.utcnow() is deprecated and scheduled for removal in a future version. Use timezone-aware objects to represent datetimes in UTC: datetime.datetime.now(datetime.UTC). I also saw a deprecation warning elsewhere for my usage of...
Here's what I learned about publishing a single file JavaScript package to NPM for my Prompts.js project. The code is in simonw/prompts-js on GitHub. The NPM package is prompts-js. A simple single file client-side package For this project, I wanted to create an old-fashioned JavaScript file that you could include in a web page using a tag. No...
My tools.simonwillison.net site is a growing collection of small HTML and JavaScript applications hosted as static files on GitHub Pages. Many of those tools take advantage of external APIs such as those provided by OpenAI and Anthropic and Google Gemini, thanks to the increasingly common access-control-allow-origin: * CORS header. I want to...
I really like Cog (previously) as a tool for automating aspects of my Python project documentation - things like the SQL schemas shown on the LLM logging page. When using cog in this way it's important to remember to run cog -r to update those generated files before pushing a commit. I've previously been enforcing this using GitHub Actions - as...
I was experimenting with wasmtime-py today and found the current documentation didn't quite give me the information that I needed. The package has a solid looking test suite, so I decided to see if I could generate additional documentation based on that. I started with a checkout of the repo: cd /tmp git clone...
If you ever run into instructions that tell you to do this: pip install flash-attn --no-build-isolation Do not try to do this. It is a trap. For some reason attempting to install this runs a compilation process which can take multiple hours. I tried to run this in Google Colab on an A100 machine that I was paying for and burned through $2 worth...
I finally figured out a process that works for me for hacking on Python CLI utilities using uv to manage my development environment, thanks to a little bit of help from Charlie Marsh. Starting a new app with cookiecutter I already have a cookiecutter template I like using for CLI applications: simonw/click-app. Thanks to uvx I don't even need to...
I ran https://simonwillison.net/ through PageSpeed Insights and it warned me that my static assets were not being served with browser caching headers: I serve static assets for my blog (mainly images) from static.simonwillison.net, which is an AWS S3 bucket served via Cloudflare. I investigated with curl -i: curl -I...
I'm still working towards adding multi-modal support to my LLM tool. In the meantime, here are notes on running prompts against images and PDFs and audio and video files from the command-line using the Google Gemini family of models. Update: I integrated the research from this TIL into my LLM tool, which can now run multi-modal prompts against...
With Claude's help I figured out what I think is the most basic version of a static site generated using Hugo. I wanted a base template that set out a common layout, then two example of pages that used that layout to render content with a custom title. This is my first time ever trying out Hugo so it's quite possible there's an even simpler...
I live in El Granada, California. Wikipedia calls us a census designated place - we don't have a mayor or city council. But we do have a Community Services District - originally responsible for our sewers, and since 2014 also responsible for our parks. And we get to vote for the board members in the upcoming November election! My partner Natalie...
I managed to get my Homebrew installation back into shape today. The first problem I was having is that it complained that macOS Sequoia was unsupported: Warning: You are using macOS 15. We do not provide support for this pre-release version. It turns out I was on an older Homebrew version. Sequoia support was added in Homebrew 4.4.0 released on...
I ran a survey on Twitter the other day to try and figure out what people mean when they use the term "agents" with respect to AI. It ended up getting over 200 replies... and then I realized that Twitter no longer shows replies to logged-out users, and the Twitter API now charges for read-access to tweets. I figured out a trick for extracting the...