Simon Willison’s Weblog

excalidraw.com (via) Really nice browser-based editor for simple diagrams using a pleasing hand-sketched style, with the ability to export them as SVG or PNG. # 30th August 2023, 4:30 pm

WebLLM supports Llama 2 70B now. The WebLLM project from MLC uses WebGPU to run large language models entirely in the browser. They recently added support for Llama 2, including Llama 2 70B, the largest and most powerful model in that family.

To my astonishment, this worked! I used a M2 Mac with 64GB of RAM and Chrome Canary and it downloaded many GBs of data... but it worked, and spat out tokens at a slow but respectable rate of 3.25 tokens/second. # 30th August 2023, 2:41 pm

Llama 2 is about as factually accurate as GPT-4 for summaries and is 30X cheaper. Anyscale offer (cheap, fast) API access to Llama 2, so they’re not an unbiased source of information—but I really hope their claim here that Llama 2 70B provides almost equivalent summarization quality to GPT-4 holds up. Summarization is one of my favourite applications of LLMs, partly because it’s key to being able to implement Retrieval Augmented Generation against your own documents—where snippets of relevant documents are fed to the model and used to answer a user’s question. Having a really high performance openly licensed summarization model is a very big deal. # 30th August 2023, 2:37 pm

Understanding Immortal Objects in Python 3.12. Abhinav Upadhyay provides a clear and detailed explanation of immortal objects coming in Python 3.12, which ensure Python no longer updates reference counts for immutable objects such as True, False, None and low-values integers. The trick (which maintains ABI compatibility) is pretty simple: a reference count value of 4294967295 now means an object is immortal, and the existing Py_INCREF and Py_DECREF macros have been updated to take that into account. # 26th August 2023, 12:08 pm

airoboros LMoE. airoboros provides a system for fine-tuning Large Language Models. The latest release adds support for LMoE—LoRA Mixture of Experts. GPT-4 is strongly rumoured to work as a mixture of experts—several (maybe 8?) 220B models each with a different specialty working together to produce the best result. This is the first open source (Apache 2) implementation of that pattern that I’ve seen. # 24th August 2023, 10:31 pm

Introducing Code Llama, a state-of-the-art large language model for coding (via) New LLMs from Meta built on top of Llama 2, in three shapes: a foundation Code Llama model, Code Llama Python that’s specialized for Python, and a Code Llama Instruct model fine-tuned for understanding natural language instructions. # 24th August 2023, 5:54 pm

llm-tracker. Leonard Lin’s constantly updated encyclopedia of all things Large Language Model: lists of models, opinions on which ones are the most useful, details for running Speech-to-Text models, code assistants and much more. # 23rd August 2023, 4:11 am

PostgreSQL Lock Conflicts (via) I absolutely love how extremely specific and niche this documentation site is. It details every single lock that PostgreSQL implements, and shows exactly which commands acquire that lock. That’s everything. I can imagine this becoming absurdly useful at extremely infrequent intervals for advanced PostgreSQL work. # 23rd August 2023, 3:08 am

Datasette Cloud and the Datasette 1.0 alphas. I sent out the Datasette Newsletter for the first time in quite a while, with updates on Datasette Cloud, the Datasette 1.0 alphas, a note about the security vulnerability in those alphas and a summary of some of my research into combining LLMs with Datasette. # 22nd August 2023, 7:56 pm

Datasette 1.0 alpha series leaks names of databases and tables to unauthenticated users. I found and fixed a security vulnerability in the Datasette 1.0 alpha series, described in this GitHub security advisory.

The vulnerability allowed unauthenticated users to see the names of the databases and tables in an otherwise private Datasette instance—though not the actual table contents.

The fix is now shipped in Datasette 1.0a4.

The vulnerability affected Datasette Cloud as well, but thankfully I was able to analyze the access logs and confirm that no unauthenticated requests had been made against any of the affected endpoints. # 22nd August 2023, 5:44 pm

Queryable Logging with Blacklite (via) Will Sargent describes how he built Blacklite, a Java library for diagnostic logging that writes log events (as zstd compressed JSON objects) to a SQLite database and maintains 5,000 entries in a “live” database while entries beyond that range are cycled out to an archive.db file, which is cycled to archive.timestamp.db when it reaches 500,000 items.

Lots of interesting notes here on using SQLite for high performance logging.

“SQLite databases are also better log files in general. Queries are faster than parsing through flat files, with all the power of SQL. A vacuumed SQLite database is only barely larger than flat file logs. They are as easy to store and transport as flat file logs, but work much better when merging out of order or interleaved data between two logs.” # 21st August 2023, 6:13 pm

Does ChatGPT have a liberal bias? (via) An excellent debunking by Arvind Narayanan and Sayash Kapoor of the “Measuring ChatGPT political bias” paper that’s been doing the rounds recently.

It turns out that paper didn’t even test ChatGPT/gpt-3.5-turbo—they ran their test against the older Da Vinci GPT3.

The prompt design was particularly flawed: they used political compass structured multiple choice: “choose between four options: strongly disagree, disagree, agree, or strongly agree”. Arvind and Sayash found that asking an open ended question was far more likely to cause the models to answer in an unbiased manner.

I liked this conclusion: “There’s a big appetite for papers that confirm users’ pre-existing beliefs [...] But we’ve also seen that chatbots’ behavior is highly sensitive to the prompt, so people can find evidence for whatever they want to believe.” # 19th August 2023, 4:53 am

Compromising LLMs: The Advent of AI Malware. The big Black Hat 2023 Prompt Injection talk, by Kai Greshake and team. The linked Whitepaper, “Not what you’ve signed up for: Compromising Real-World LLM-Integrated Applications with Indirect Prompt Injection”, is the most thorough review of prompt injection attacks I’ve seen yet. # 18th August 2023, 2:46 am

Running my own LLM (via) Nelson Minar describes running LLMs on his own computer using my LLM tool and llm-gpt4all plugin, plus some notes on trying out some of the other plugins. # 16th August 2023, 10:42 pm

An Iowa school district is using ChatGPT to decide which books to ban. I’m quoted in this piece by Benj Edwards about an Iowa school district that responded to a law requiring books be removed from school libraries that include “descriptions or visual depictions of a sex act” by asking ChatGPT “Does [book] contain a description or depiction of a sex act?”.

I talk about how this is the kind of prompt that frequent LLM users will instantly spot as being unlikely to produce reliable results, partly because of the lack of transparency from OpenAI regarding the training data that goes into their models. If the models haven’t seen the full text of the books in question, how could they possibly provide a useful answer? # 16th August 2023, 10:33 pm

Introducing datasette-write-ui: a Datasette plugin for editing, inserting, and deleting rows. Alex García is working with me on Datasette Cloud for the next few months, graciously sponsored by Fly. We will be working in public, releasing open source code and documenting how to build a multi-tenant SaaS product using Fly Machines.

Alex’s first project is datasette-write-ui, a plugin that finally lets you directly edit data stored inside Datasette. Alex wrote about the plugin on our new Datasette Cloud blog. # 16th August 2023, 1:48 am

Welcome to Datasette Cloud. We launched the Datasette Cloud blog today! The SaaS hosted version of Datasette is ready to start onboarding more users—this post describes what it can do so far and hints at what’s planned to come next. # 16th August 2023, 1:46 am

Write about what you learn. It pushes you to understand topics better. (via) Addy Osmani clearly articulates why writing frequently is such a powerful tool for learning more effectively. This post doesn’t mention TILs but it perfectly encapsulates the value I get from publishing them. # 14th August 2023, 2:50 pm

Lark parsing library JSON tutorial (via) A very convincing tutorial for a new-to-me parsing library for Python called Lark.

The tutorial covers building a full JSON parser from scratch, which ends up being just 19 lines of grammar definition code and 15 lines for the transformer to turn that tree into the final JSON.

It then gets into the details of optimization—the default Earley algorithm is quite slow, but swapping that out for a LALR parser (a one-line change) provides a 5x speedup for this particular example. # 13th August 2023, 9:50 pm

deno_python (via) A wildly impressive hack: deno_python uses Deno’s FFI interface to load your system’s Python framework (.dll/.dylib/.so) and sets up JavaScript proxy objects for imported Python objects—so you can run JavaScript code that instantiates objects from Python libraries and uses them to process data in different ways.

The latest release added pip support, so things like ’const np = await pip.import(“numpy”)’ now work. # 12th August 2023, 10:14 pm

llm-mlc (via) My latest plugin for LLM adds support for models that use the MLC Python library—which is the first library I’ve managed to get to run Llama 2 with GPU acceleration on my M2 Mac laptop. # 12th August 2023, 5:33 am

Dependency Management Data (via) This is a really neat CLI tool by Jamie Tanna, built using Go and SQLite but with a feature that embeds a Datasette instance (literally shelling out to start the process running from within the Go application) to provide an interface for browsing the resulting database.

It addresses the challenge of keeping track of the dependencies used across an organization, by gathering them into a SQLite database from a variety of different sources—currently Dependabot, Renovate and some custom AWS tooling.

The “Example” page links to a live Datasette instance and includes video demos of the tool in action. # 11th August 2023, 3:54 pm

Shamir Secret Sharing (via) Cracking war story from Max Levchin about the early years of PayPal, in which he introduces an implementation of Shamir Secret Sharing to encrypt their master payment credential table... and then finds that the 3-of-8 passwords needed to decrypt it and bring the site back online don’t appear to work. # 11th August 2023, 3:48 pm

Getting creative with embeddings (via) Amelia Wattenberger describes a neat application of embeddings I haven’t seen before: she wanted to build a system that could classify individual sentences in terms of how “concrete” or “abstract” they are. So she generated several example sentences for each of those categories, embedded then and calculated the average of those embeddings.

And now she can get a score for how abstract vs concrete a new sentence is by calculating its embedding and seeing where it falls in the 1500 dimension space between those two other points. # 10th August 2023, 7:05 pm

Datasette 1.0a3. A new Datasette alpha release. This one previews the new default JSON API design that’s coming in 1.0—the single most significant change in the 1.0 milestone, since I plan to keep that API stable for many years to come. # 9th August 2023, 8:49 pm

Llama from scratch (or how to implement a paper without crying) (via) Brian Kitano implemented the model described in the Llama paper against TinyShakespeare, from scratch, using Python and PyTorch. This write-up is fantastic—meticulous, detailed and deeply informative. It would take several hours to fully absorb and follow everything Brian does here but it would provide multiple valuable lessons in understanding how all of this stuff fits together. # 9th August 2023, 7:21 pm

Python cocktail: mix a context manager and an iterator in equal parts (via) Explanation of a neat trick used by the Tenacity Python library, which provides a mechanism for retrying a chunk of code automatically on errors up to three times using a mixture of an iterator and a context manager to work around Python’s lack of multi-line lambda functions. # 6th August 2023, 5:44 pm

You can stop using user-scalable=no and maximum-scale=1 in viewport meta tags now. Luke Plant points out that your meta viewport tag should stick to just “width=device-width, initial-scale=1” these days—the user-scalable=no and maximum-scale=1 attributes are no longer necessary, and have a negative impact on accessibility, especially for Android users. # 4th August 2023, 11:41 pm