45 items tagged “training-data”
Data used to train LLMs and other machine learning models.
2023
I’ve resigned from my role leading the Audio team at Stability AI, because I don’t agree with the company’s opinion that training generative AI models on copyrighted works is ‘fair use’.
[...] I disagree because one of the factors affecting whether the act of copying is fair use, according to Congress, is “the effect of the use upon the potential market for or value of the copyrighted work”. Today’s generative AI models can clearly be used to create works that compete with the copyrighted works they are trained on. So I don’t see how using copyrighted works to train generative AI models of this nature can be considered fair use.
But setting aside the fair use argument for a moment — since ‘fair use’ wasn’t designed with generative AI in mind — training generative AI models in this way is, to me, wrong. Companies worth billions of dollars are, without permission, training generative AI models on creators’ works, which are then being used to create new content that in many cases can compete with the original works.
It’s infuriatingly hard to understand how closed models train on their input
One of the most common concerns I see about large language models regards their training data. People are worried that anything they say to ChatGPT could be memorized by it and spat out to other users. People are concerned that anything they store in a private repository on GitHub might be used as training data for future versions of Copilot.
[... 1,465 words]Introducing speech-to-text, text-to-speech, and more for 1,100+ languages (via) New from Meta AI: Massively Multilingual Speech. “MMS supports speech-to-text and text-to-speech for 1,107 languages and language identification for over 4,000 languages. [...] Some of these, such as the Tatuyo language, have only a few hundred speakers, and for most of these languages, no prior speech technology exists.”
It’s licensed CC-BY-NC 4.0 though, so it’s not available for commercial use.
“In a like-for-like comparison with OpenAI’s Whisper, we found that models trained on the Massively Multilingual Speech data achieve half the word error rate, but Massively Multilingual Speech covers 11 times more languages.”
The training data was mostly sourced from audio Bible translations.
Inside the secret list of websites that make AI chatbots sound smart. Washington Post story digging into the C4 dataset—Colossal Clean Crawled Corpus, a filtered version of Common Crawl that’s often used for training large language models. They include a neat interactive tool for searching a domain to see if it’s included—TIL that simonwillison.net is the 106,649th ranked site in C4 by number of tokens, 189,767 total—0.0001% of the total token volume in C4.
What’s in the RedPajama-Data-1T LLM training set
RedPajama is “a project to create leading open-source models, starts by reproducing LLaMA training dataset of over 1.2 trillion tokens”. It’s a collaboration between Together, Ontocord.ai, ETH DS3Lab, Stanford CRFM, Hazy Research, and MILA Québec AI Institute.
[... 1,077 words]RedPajama, a project to create leading open-source models, starts by reproducing LLaMA training dataset of over 1.2 trillion tokens. With the amount of projects that have used LLaMA as a foundation model since its release two months ago—despite its non-commercial license—it’s clear that there is a strong desire for a fully openly licensed alternative.
RedPajama is a collaboration between Together, Ontocord.ai, ETH DS3Lab, Stanford CRFM, Hazy Research, and MILA Québec AI Institute aiming to build exactly that.
Step one is gathering the training data: the LLaMA paper described a 1.2 trillion token training set gathered from sources that included Wikipedia, Common Crawl, GitHub, arXiv, Stack Exchange and more.
RedPajama-Data-1T is an attempt at recreating that training set. It’s now available to download, as 2,084 separate multi-GB jsonl files—2.67TB total.
Even without a trained model, this is a hugely influential contribution to the world of open source LLMs. Any team looking to build their own LLaMA from scratch can now jump straight to the next stage, training the model.
Replacing my best friends with an LLM trained on 500,000 group chat messages (via) Izzy Miller used a 7 year long group text conversation with five friends from college to fine-tune LLaMA, such that it could simulate ongoing conversations. They started by extracting the messages from the iMessage SQLite database on their Mac, then generated a new training set from those messages and ran it using code from the Stanford Alpaca repository. This is genuinely one of the clearest explanations of the process of fine-tuning a model like this I’ve seen anywhere.
ROOTS search tool (via) BLOOM is one of the most interesting completely openly licensed language models. The ROOTS corpus is the training data that was collected for it, and this tool lets you run searches directly against that corpus. I tried searching for my own name and got an interesting insight into what it knows about me.
mitsua-diffusion-one (via) “Mitsua Diffusion One is a latent text-to-image diffusion model, which is a successor of Mitsua Diffusion CC0. This model is trained from scratch using only public domain/CC0 or copyright images with permission for use.” I’ve been talking about how much I’d like to try out a “vegan” AI model trained entirely on out-of-copyright images for ages, and here one is! It looks like the training data mainly came from CC0 art gallery collections such as the Metropolitan Museum of Art Open Access.
Don’t trust AI to talk accurately about itself: Bard wasn’t trained on Gmail
Earlier this month I wrote about how ChatGPT can’t access the internet, even though it really looks like it can. Consider this part two in the series. Here’s another common and non-intuitive mistake people make when interacting with large language model AI systems: asking them questions about themselves.
[... 1,950 words]Adobe made an AI image generator — and says it didn’t steal artists’ work to do it. Adobe Firefly is a brand new text-to-image model which Adobe claim was trained entirely on fully licensed imagery—either out of copyright, specially licensed or part of the existing Adobe Stock library. I’m sure they have the license, but I still wouldn’t be surprised to hear complaints from artists who licensed their content to Adobe Stock who didn’t anticipate it being used for model training.
Exploring MusicCaps, the evaluation data released to accompany Google’s MusicLM text-to-music model
Google Research just released MusicLM: Generating Music From Text. It’s a new generative AI model that takes a descriptive prompt and produces a “high-fidelity” music track. Here’s the paper (and a more readable version using arXiv Vanity).
[... 1,323 words]2022
Exploring 10m scraped Shutterstock videos used to train Meta’s Make-A-Video text-to-video model
Make-A-Video is a new “state-of-the-art AI system that generates videos from text” from Meta AI. It looks incredible—it really is DALL-E / Stable Diffusion for video. And it appears to have been trained on 10m video preview clips scraped from Shutterstock.
[... 923 words]Exploring the training data behind Stable Diffusion
Two weeks ago, the Stable Diffusion image generation model was released to the public. I wrote about this last week, in Stable Diffusion is a really big deal—a post which has since become one of the top ten results for “stable diffusion” on Google and shown up in all sorts of different places online.
[... 2,897 words]Exploring 12 Million of the 2.3 Billion Images Used to Train Stable Diffusion’s Image Generator. Andy Baio and I collaborated on an investigation into the training set used for Stable Diffusion. I built a Datasette instance with 12m image records sourced from the LAION-Aesthetics v2 6+ aesthetic score data used as part of the training process, and built a tool so people could run searches and explore the data. Andy did some extensive analysis of things like the domains scraped for the images and names of celebrities and artists represented in the data. His write-up here explains our project in detail and some of the patterns we’ve uncovered so far.