Blogmarks
Filters: Sorted by date
You can now train a 70b language model at home (via) Jeremy Howard and team: “Today, we’re releasing Answer.AI’s first project: a fully open source system that, for the first time, can efficiently train a 70b large language model on a regular desktop computer with two or more standard gaming GPUs (RTX 3090 or 4090).”
This is about fine-tuning an existing model, not necessarily training one from scratch.
There are two tricks at play here. The first is QLoRA, which can be used to train quantized models despite the reduced precision usually preventing gradient descent from working correctly.
QLoRA can bring the memory requirements for a 70b model down to 35GB, but gaming GPUs aren’t quite that big. The second trick is Meta’s Fully Sharded Data Parallel or FSDP library, which can shard a model across GPUs. Two consumer 24GB GPUs can then handle the 70b training run.
Become a Wikipedian in 30 minutes (via) A characteristically informative and thoughtful guide to getting started with Wikipedia editing by Molly White - video accompanied by a full transcript.
I found the explanation of Reliable Sources particularly helpful, including why Wikipedia prefers secondary to primary sources.
The way we determine reliability is typically based on the reputation for editorial oversight, and for factchecking and corrections. For example, if you have a reference book that is published by a reputable publisher that has an editorial board and that has edited the book for accuracy, if you know of a newspaper that has, again, an editorial team that is reviewing articles and issuing corrections if there are any errors, those are probably reliable sources.
Eloquent JavaScript, 4th edition (2024) (via) Marijn Haverbeke is the creator of both the CodeMirror JavaScript code editor library (used by Datasette and many other projects) and the ProseMirror rich-text editor. Eloquent JavaScript is his Creative Commons licensed book on JavaScript, first released in 2007 and now in its 4th edition.
I’ve only dipped into it myself but it has an excellent reputation.
Inflection-2.5: meet the world’s best personal AI (via) I’ve not been paying much attention to Inflection’s Pi since it released last year, but yesterday they released a new version that they claim is competitive with GPT-4.
“Inflection-2.5 approaches GPT-4’s performance, but used only 40% of the amount of compute for training.”
(I wasn’t aware that the compute used to train GPT-4 was public knowledge.)
If this holds true, that means that the GPT-4 barrier has been well and truly smashed: we now have Claude 3 Opus, Gemini 1.5, Mistral Large and Inflection-2.5 in the same class as GPT-4, up from zero contenders just a month ago.
American Community Survey Data via FTP. I got talking to some people from the US Census at NICAR today and asked them if there was a way to download their data in bulk (in addition to their various APIs)... and there was!
I had heard of the American Community Survey but I hadn’t realized that it’s gathered on a yearly basis, as a 5% sample compared to the full every-ten-years census. It’s only been running for ten years, and there’s around a year long lead time on the survey becoming available.
Training great LLMs entirely from ground zero in the wilderness as a startup. Yi Tay has a really interesting perspective on training LLMs, having worked at Google Brain before co-founding an independent startup, Reka.
At Google the clusters are provided for you. On the outside, Yi finds himself bargaining for cluster resources from a wide range of vendors—and running into enormous variance in quality.
“We’ve seen clusters that range from passable (just annoying problems that are solvable with some minor SWE hours) to totally unusable clusters that fail every few hours due to a myriad of reasons.”
The Claude 3 system prompt, explained. Anthropic research scientist Amanda Askell provides a detailed breakdown of the Claude 3 system prompt in a Twitter thread.
This is some fascinating prompt engineering. It's also great to see an LLM provider proudly documenting their system prompt, rather than treating it as a hidden implementation detail.
The prompt is pretty succinct. The three most interesting paragraphs:
If it is asked to assist with tasks involving the expression of views held by a significant number of people, Claude provides assistance with the task even if it personally disagrees with the views being expressed, but follows this with a discussion of broader perspectives.
Claude doesn't engage in stereotyping, including the negative stereotyping of majority groups.
If asked about controversial topics, Claude tries to provide careful thoughts and objective information without downplaying its harmful content or implying that there are reasonable perspectives on both sides.
How I use git worktrees (via) TIL about worktrees, a Git feature that lets you have multiple repository branches checked out to separate directories at the same time.
The default UI for them is a little unergonomic (classic Git) but Bill Mill here shares a neat utility script for managing them in a more convenient way.
One particularly neat trick: Bill’s “worktree” Bash script checks for a node_modules folder and, if one exists, duplicates it to the new directory using copy-on-write, saving you from having to run yet another lengthy “npm install”.
Wikimedia Commons Category:Bach Dancing & Dynamite Society. After creating a new Wikipedia page for the Bach Dancing & Dynamite Society in Half Moon Bay I ran a search across Wikipedia for other mentions of the venue... and found 41 artist pages that mentioned it in a photo caption.
On further exploration it turns out that Brian McMillen, the official photographer for the venue, has been uploading photographs to Wikimedia Commons since 2007 and adding them to different artist pages. Brian has been a jazz photographer based out of Half Moon Bay for 47 years and has an amazing portfolio of images. It’s thrilling to see him share them on Wikipedia in this way.
Observable Framework 1.1 (via) Less than three weeks after 1.0, the 1.1 release adds a whole lot of interesting new stuff. The signature feature is self-hosted npm imports: Framework 1.0 linked out to CDN hosted copies of libraries, but 1.1 fetches copies locally and then bundles that code with the deployed static site.
This works by using the acorn JavaScript parsing library to statically analyze the code and find all of the relevant imports.
Wikipedia: Bach Dancing & Dynamite Society (via) I created my first Wikipedia page! The Bach Dancing & Dynamite Society is a really neat live music venue in Half Moon Bay which has been showcasing world-class jazz talent for over 50 years. I attended a concert there for the first time on Sunday and was surprised to see it didn’t have a page yet.
Creating a Wikipedia page is an interesting process. New pages on English Wikipedia created by infrequent editors stay in “draft” mode until they’ve been approved by a member of “WikiProject Articles for creation”—the standards are really high, especially around sources of citations. I spent quite a while tracking down good citation references for the key facts I used in my first draft for the page.
llm-claude-3. I built a new plugin for LLM—my command-line tool and Python library for interacting with Large Language Models—which adds support for the new Claude 3 models from Anthropic.
The new Claude 3 model family from Anthropic. Claude 3 is out, and comes in three sizes: Opus (the largest), Sonnet and Haiku.
Claude 3 Opus has self-reported benchmark scores that consistently beat GPT-4. This is a really big deal: in the 12+ months since the GPT-4 release no other model has consistently beat it in this way. It’s exciting to finally see that milestone reached by another research group.
The pricing model here is also really interesting. Prices here are per-million-input-tokens / per-million-output-tokens:
Claude 3 Opus: $15 / $75
Claude 3 Sonnet: $3 / $15
Claude 3 Haiku: $0.25 / $1.25
All three models have a 200,000 length context window and support image input in addition to text.
Compare with today’s OpenAI prices:
GPT-4 Turbo (128K): $10 / $30
GPT-4 8K: $30 / $60
GPT-4 32K: $60 / $120
GPT-3.5 Turbo: $0.50 / $1.50
So Opus pricing is comparable with GPT-4, more than GPT-4 Turbo and significantly cheaper than GPT-4 32K... Sonnet is cheaper than all of the GPT-4 models (including GPT-4 Turbo), and Haiku (which has not yet been released to the Claude API) will be cheaper even than GPT-3.5 Turbo.
It will be interesting to see if OpenAI respond with their own price reductions.
Who Am I? Conditional Prompt Injection Attacks with Microsoft Copilot (via) New prompt injection variant from Johann Rehberger, demonstrated against Microsoft Copilot. If the LLM tool you are interacting with has awareness of the identity of the current user you can create targeted prompt injection attacks which only activate when an exploit makes it into the token context of a specific individual.
The One Billion Row Challenge in Go: from 1m45s to 4s in nine solutions (via) How fast can you read a billion semicolon delimited (name;float) lines and output a min/max/mean summary for each distinct name—13GB total?
Ben Hoyt describes his 9 incrementally improved versions written in Go in detail. The key optimizations involved custom hashmaps, optimized line parsing and splitting the work across multiple CPU cores.
The Radio Squirrels of Point Reyes (via) Beautiful photo essay by Ann Hermes about the band of volunteer “radio squirrels” keeping maritime morse code radio transmissions alive in the Point Reyes National Seashore.
Streaming HTML out of order without JavaScript (via) A really interesting new browser capability. If you serve the following HTML:
<template shadowrootmode="open">
<slot name="item-1">Loading...</slot>
</template>
Then later in the same page stream an element specifying that slot:
<span slot="item-1">Item number 1</span>
The previous slot will be replaced while the page continues to load.
I tried the demo in the most recent Chrome, Safari and Firefox (and Mobile Safari) and it worked in all of them.
The key feature is shadowrootmode=open, which looks like it was added to Firefox 123 on February 19th 2024 - the other two browsers are listed on caniuse.com as gaining it around March last year.
Endatabas (via) Endatabas is “an open source immutable database”—also described as “SQL document database with full history”.
It uses a variant of SQL which allows you to insert data into tables that don’t exist yet (they’ll be created automatically) then run standard select queries, joins etc. It maintains a full history of every record and supports the recent SQL standard “FOR SYSTEM_TIME AS OF” clause for retrieving historical records as they existed at a specified time (it defaults to the most recent versions).
It’s written in Common Lisp plus a bit of Rust, and includes Docker images for running the server and client libraries in JavaScript and Python. The on-disk storage format is Apache Arrow, the license is AGPL and it’s been under development for just over a year.
It’s also a document database: you can insert JSON-style nested objects directly into a table, and query them with path expressions like “select users.friends[1] from users where id = 123;”
They have a WebAssembly version and a nice getting started tutorial which you can try out directly in your browser.
Their “Why?” page lists full history, time travel queries, separation of storage from compute, schemaless tables and columnar storage as the five pillars that make up their product. I think it’s a really interesting amalgamation of ideas.
Datasette 1.0a12. Another alpha release, this time with a new query_actions() plugin hook, a new design for the table, database and query actions menus, a “does not contain” table filter and a fix for a minor bug with the JavaScript makeColumnActions() plugin mechanism.
GGUF, the long way around (via) Vicki Boykis dives deep into the GGUF format used by llama.cpp, after starting with a detailed description of how PyTorch models work and how they are traditionally persisted using Python pickle.
Pickle lead to safetensors, a format that avoided the security problems with downloading and running untrusted pickle files.
Llama.cpp introduced GGML, which popularized 16-bit (as opposed to 32-bit) quantization and bundled metadata and tensor data in a single file.
GGUF fixed some design flaws in GGML and is the default format used by Llama.cpp today.
The Zen of Python, Unix, and LLMs. Here’s the YouTube recording of my 1.5 hour conversation with Hugo Bowne-Anderson yesterday.
I fed a Whisper transcript to Google Gemini Pro 1.5 and asked it for the themes from our conversation, and it said we talked about “Python’s success and versatility, the rise and potential of LLMs, data sharing and ethics in the age of LLMs, Unix philosophy and its influence on software development and the future of programming and human-computer interaction”.
Testcontainers (via) Not sure how I missed this: Testcontainers is a family of testing libraries (for Python, Go, JavaScript, Ruby, Rust and a bunch more) that make it trivial to spin up a service such as PostgreSQL or Redis in a container for the duration of your tests and then spin it back down again.
The Python example code is delightful:
redis = DockerContainer("redis:5.0.3-alpine").with_exposed_ports(6379)
redis.start()
wait_for_logs(redis, "Ready to accept connections")
I much prefer integration-style tests over unit tests, and I like to make sure any of my projects that depend on PostgreSQL or similar can run their tests against a real running instance. I've invested heavily in spinning up Varnish or Elasticsearch ephemeral instances in the past - Testcontainers look like they could save me a lot of time.
The open source project started in 2015, span off a company called AtomicJar in 2021 and was acquired by Docker in December 2023.
The Zen of Python, Unix, and LLMs with Simon Willison (via) I’m participating in a live online fireside chat with Hugo Bowne-Anderson tomorrow afternoon (3pm Pacific / 6pm Eastern / 11pm GMT) talking about LLMs, Datasette, my open source process, applying the Unix pipes philosophy to LLMs and a whole lot more. It’s free to register.
All you need is Wide Events, not “Metrics, Logs and Traces” (via) I’ve heard great things about Meta’s internal observability platform Scuba, here’s an explanation from ex-Meta engineer Ivan Burmistrov describing the value it provides and comparing it to the widely used OpenTelemetry stack.
Mistral Large. Mistral Medium only came out two months ago, and now it's followed by Mistral Large. Like Medium, this new model is currently only available via their API. It scores well on benchmarks (though not quite as well as GPT-4) but the really exciting feature is function support, clearly based on OpenAI's own function design.
Functions are now supported via the Mistral API for both Mistral Large and the new Mistral Small, described as follows:
Mistral Small, optimised for latency and cost. Mistral Small outperforms Mixtral 8x7B and has lower latency, which makes it a refined intermediary solution between our open-weight offering and our flagship model.
dclient 0.3. dclient is my CLI utility for working with remote Datasette instances—in particular for authenticating with them and then running both read-only SQL queries and inserting data using the new Datasette write JSON API. I just picked up work on the project again after a six month gap—the insert command can now be used to constantly stream data directly to hosted Datasette instances such as Datasette Cloud.
Upside down table trick with CSS (via) I was complaining how hard it is to build a horizontally scrollable table with a scrollbar at the top rather than the bottom and RGBCube on Lobste.rs suggested rotating the container 180 degrees and then the table contents and headers 180 back again... and it totally works! Demo in this CodePen.
How to make self-hosted maps that work everywhere and cost next to nothing. Chris Amico provides a detailed roundup of the state of web mapping in 2024. It’s never been easier to entirely host your own mapping infrastructure, thanks to OpenStreetMap, Overture, MBTiles, PMTiles, Maplibre and a whole ecosystem of other fine open source projects.
I like Protomaps creator Brandon Liu’s description of this: “post-scarcity web mapping”.
Does Offering ChatGPT a Tip Cause it to Generate Better Text? An Analysis (via) Max Woolf:“I have a strong hunch that tipping does in fact work to improve the output quality of LLMs and its conformance to constraints, but it’s very hard to prove objectively. [...] Let’s do a more statistical, data-driven approach to finally resolve the debate.”
Bloom Filters, explained by Sam Rose. Beautifully designed explanation of bloom filters, complete with interactive demos that illustrate exactly how they work.