773 items tagged “llms”
Large Language Models (LLMs) are the class of technology behind generative text AI systems like OpenAI's ChatGPT, Google's Gemini and Anthropic's Claude.
2023
The largest model in the PaLM 2 family, PaLM 2-L, is significantly smaller than the largest PaLM model but uses more training compute. Our evaluation results show that PaLM 2 models significantly outperform PaLM on a variety of tasks, including natural language generation, translation, and reasoning. These results suggest that model scaling is not the only way to improve performance. Instead, performance can be unlocked by meticulous data selection and efficient architecture/objectives. Moreover, a smaller but higher quality model significantly improves inference efficiency, reduces serving cost, and enables the model’s downstream application for more applications and users.
— PaLM 2 Technical Report, PDF
Language models can explain neurons in language models (via) Fascinating interactive paper by OpenAI, describing how they used GPT-4 to analyze the concepts tracked by individual neurons in their much older GPT-2 model. “We generated cluster labels by embedding each neuron explanation using the OpenAI Embeddings API, then clustering them and asking GPT-4 to label each cluster.”
Jsonformer: A Bulletproof Way to Generate Structured JSON from Language Models. This is such an interesting trick. A common challenge with LLMs is getting them to output a specific JSON shape of data reliably, without occasionally messing up and generating invalid JSON or outputting other text.
Jsonformer addresses this in a truly ingenious way: it implements code that interacts with the logic that decides which token to output next, influenced by a JSON schema. If that code knows that the next token after a double quote should be a comma it can force the issue for that specific token.
This means you can get reliable, robust JSON output even for much smaller, less capable language models.
It’s built against Hugging Face transformers, but there’s no reason the same idea couldn’t be applied in other contexts as well.
Introducing MPT-7B: A New Standard for Open-Source, Commercially Usable LLMs (via) There’s a lot to absorb about this one. Mosaic trained this model from scratch on 1 trillion tokens, at a cost of $200,000 taking 9.5 days. It’s Apache-2.0 licensed and the model weights are available today.
They’re accompanying the base model with an instruction-tuned model called MPT-7B-Instruct (licensed for commercial use) and a non-commercially licensed MPT-7B-Chat trained using OpenAI data. They also announced MPT-7B-StoryWriter-65k+—“a model designed to read and write stories with super long context lengths”—with a previously unheard of 65,000 token context length.
They’re releasing these models mainly to demonstrate how inexpensive and powerful their custom model training service is. It’s a very convincing demo!
No Moat: Closed AI gets its Open Source wakeup call — ft. Simon Willison (via) I joined the Latent Space podcast yesterday (on short notice, so I was out and about on my phone) to talk about the leaked Google memo about open source LLMs. This was a Twitter Space, but swyx did an excellent job of cleaning up the audio and turning it into a podcast.
Leaked Google document: “We Have No Moat, And Neither Does OpenAI”
SemiAnalysis published something of a bombshell leaked document this morning: Google “We Have No Moat, And Neither Does OpenAI”.
[... 1,073 words]OpenLLaMA. The first openly licensed model I’ve seen trained on the RedPajama dataset. This initial release is a 7B model trained on 200 billion tokens, but the team behind it are promising a full 1 trillion token model in the near future. I haven’t found a live demo of this one running anywhere yet.
replit-code-v1-3b (via) As promised last week, Replit have released their 2.7b “Causal Language Model”, a foundation model trained from scratch in partnership with MosaicML with a focus on code completion. It’s licensed CC BY-SA-4.0 and is available for commercial use. They repo includes a live demo and initial experiments with it look good—you could absolutely run a local GitHub Copilot style editor on top of this model.
We show for the first time that large-scale generative pretrained transformer (GPT) family models can be pruned to at least 50% sparsity in one-shot, without any retraining, at minimal loss of accuracy. [...] We can execute SparseGPT on the largest available open-source models, OPT-175B and BLOOM-176B, in under 4.5 hours, and can reach 60% unstructured sparsity with negligible increase in perplexity: remarkably, more than 100 billion weights from these models can be ignored at inference time.
— SparseGPT, by Elias Frantar and Dan Alistarh
Prompt injection explained, with video, slides, and a transcript
I participated in a webinar this morning about prompt injection, organized by LangChain and hosted by Harrison Chase, with Willem Pienaar, Kojin Oshiba (Robust Intelligence), and Jonathan Cohen and Christopher Parisien (Nvidia Research).
[... 3,120 words]Let’s be bear or bunny
The Machine Learning Compilation group (MLC) are my favourite team of AI researchers at the moment.
[... 599 words]Enriching data with GPT3.5 and SQLite SQL functions
I shipped openai-to-sqlite 0.3 yesterday with a fun new feature: you can now use the command-line tool to enrich data in a SQLite database by running values through an OpenAI model and saving the results, all in a single SQL query.
[... 1,219 words]MLC LLM (via) From MLC, the team that gave us Web LLM and Web Stable Diffusion. “MLC LLM is a universal solution that allows any language model to be deployed natively on a diverse set of hardware backends and native applications”. I installed their iPhone demo from TestFlight this morning and it does indeed provide an offline LLM that runs on my phone. It’s reasonably capable—the underlying model for the app is vicuna-v1-7b, a LLaMA derivative.
GPT-3 token encoder and decoder. I built an Observable notebook with an interface to encode, decode and search through GPT-3 tokens, building on top of a notebook by EJ Fox and Ian Johnson.
How prompt injection attacks hijack today’s top-end AI – and it’s really tough to fix. Thomas Claburn interviewed me about prompt injection for the Register. Lots of direct quotes from our phone call in here—we went pretty deep into why it’s such a difficult problem to address.
The Dual LLM pattern for building AI assistants that can resist prompt injection
I really want an AI assistant: a Large Language Model powered chatbot that can answer questions and perform actions for me based on access to my private data and tools.
[... 2,547 words]A lot of people who claim to be doing prompt engineering today are actually just blind prompting. "Blind Prompting" is a term I am using to describe the method of creating prompts with a crude trial-and-error approach paired with minimal or no testing and a very surface level knowedge of prompting. Blind prompting is not prompt engineering. [...] In this blog post, I will make the argument that prompt engineering is a real skill that can be developed based on real experimental methodologies.
Bard now helps you code (via) Google have enabled Bard’s code generation abilities—these were previously only available through jailbreaking. It’s pretty good—I got it to write me code to download a CSV file and insert it into a SQLite database—though when I challenged it to protect against SQL injection it hallucinated a non-existent “cursor.prepare()” method. Generated code can be exported to a Colab notebook with a click.
Stability AI Launches the First of its StableLM Suite of Language Models (via) 3B and 7B base models, with 15B and 30B are on the way. CC BY-SA-4.0. “StableLM is trained on a new experimental dataset built on The Pile, but three times larger with 1.5 trillion tokens of content. We will release details on the dataset in due course.”
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.
LLaVA: Large Language and Vision Assistant (via) Yet another multi-modal model combining a vision model (pre-trained CLIP ViT-L/14) and a LLaMA derivative model (Vicuna). The results I get from their demo are even more impressive than MiniGPT-4. Also includes a new training dataset, LLaVA-Instruct-150K, derived from GPT-4 and subject to the same warnings about the OpenAI terms of service.
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.
MiniGPT-4 (via) An incredible project with a poorly chosen name. A team from King Abdullah University of Science and Technology in Saudi Arabia combined Vicuna-13B (a model fine-tuned on top of Facebook’s LLaMA) with the BLIP-2 vision-language model to create a model that can conduct ChatGPT-style conversations around an uploaded image. The demo is very impressive, and the weights are available to download—45MB for MiniGPT-4, but you’ll need the much larger Vicuna and LLaMA weights as well.
Web LLM runs the vicuna-7b Large Language Model entirely in your browser, and it’s very impressive
A month ago I asked Could you train a ChatGPT-beating model for $85,000 and run it in a browser?. $85,000 was a hypothetical training cost for LLaMA 7B plus Stanford Alpaca. “Run it in a browser” was based on the fact that Web Stable Diffusion runs a 1.9GB Stable Diffusion model in a browser, so maybe it’s not such a big leap to run a small Large Language Model there as well.
[... 2,276 words]Although fine-tuning can feel like the more natural option—training on data is how GPT learned all of its other knowledge, after all—we generally do not recommend it as a way to teach the model knowledge. Fine-tuning is better suited to teaching specialized tasks or styles, and is less reliable for factual recall. [...] In contrast, message inputs are like short-term memory. When you insert knowledge into a message, it's like taking an exam with open notes. With notes in hand, the model is more likely to arrive at correct answers.
— Ted Sanders, OpenAI
New prompt injection attack on ChatGPT web version. Markdown images can steal your chat data. An ingenious new prompt injection / data exfiltration vector from Roman Samoilenko, based on the observation that ChatGPT can render markdown images in a way that can exfiltrate data to the image hosting server by embedding it in the image URL. Roman uses a single pixel image for that, and combines it with a trick where copy events on a website are intercepted and prompt injection instructions are appended to the copied text, in order to trick the user into pasting the injection attack directly into ChatGPT.
Update: They finally started mitigating this in December 2023.
One way to avoid unspotted prediction errors is for the technology in its current state to have early and frequent contact with reality as it is iteratively developed, tested, deployed, and all the while improved. And there are creative ideas people don’t often discuss which can improve the safety landscape in surprising ways — for example, it’s easy to create a continuum of incrementally-better AIs (such as by deploying subsequent checkpoints of a given training run), which presents a safety opportunity very unlike our historical approach of infrequent major model upgrades.
Prompt injection: What’s the worst that can happen?
Activity around building sophisticated applications on top of LLMs (Large Language Models) such as GPT-3/4/ChatGPT/etc is growing like wildfire right now.
[... 2,302 words]Building LLM applications for production. Chip Huyen provides a useful, in-depth review of the challenges involved in taking an app built on top of a LLM from prototype to production, including issues such as prompt ambiguity and unpredictability, cost and latency concerns, challenges in testing and updating to new models. She also lists some promising use-cases she’s seeing for categories of application built on these tools.