151
Creating Shazam in Java (2010)
A couple of days ago I encountered this article: How Shazam WorksThis got me interested in how a program like Shazam works… And more importantly, how hard is...
New
The latest blog posts to take off on Hacker News. 301 posts.
152
Emergent Features in 6.7B+ Param Transformers
When I attended NAACL, I wanted to do a little test. I had two pitches for my LLM.int8() paper. One pitch is about how I use advanced quantization methods to achieve no performance degradation transformer inference at scale that makes large models more accessible. The other pitch talks about emergent outliers in…
153
154
Adventure game graphics with DALL-E 2
Is DALL-E 2 fit for use in creative work? A case study involving pixel graphics for a point-and-click adventure game.
155
The future of interactive theorem proving?
This is a guest post, written by Zhangir Azerbayev. Zhangir is an undergraduate at Yale, majoring in computer science and mathematics. He completed this work while visiting Carnegie Mellon’s …
156
The AI Art Apocalypse
This image was created by an AI, MidJourney. All I had to do was type in a prompt (“wildfire”) and aspect ratio. This AI is pretty good, but nowhere near the state of the art, and AI like it are, over the next few years, going to make art like this available within seconds at […]
157
How to create a blog post title optimizer with GPT-3 and Hacker News data
GPT-3 says the title for this post is very bad.
158
Copyright infringement in AI art
As AI creative tools are becoming widespread, the question of copyright of AI creations has also taken centre-stage. But while copyright nerds obsess over the authorship question, the issue that is…
159
Stable Diffusion launch announcement
The model itself builds upon the work of the team at CompVis and Runway in their widely used latent diffusion model combined with insights from the conditional diffusion models by our lead generative AI developer Katherine Crowson, Dall-E 2 by Open AI, Imagen by Google Brain and many others. We are
160
I replaced all our blog thumbnails using DALL·E 2
An infinite number of machine monkeys paint an infinite number of blog thumbnails
161
162
Run your own DALL-E-like image generator
Run your own text to image prompts with CUDA, a bunch of disk space, and an insane amount of memory.
163
BlenderBot 3: A 175B parameter, publicly available chatbot
We’re announcing that Meta AI has built and released BlenderBot 3, the first 175B parameter, publicly available chatbot complete with model weights, code, datasets, and model cards. We’ve deployed it in a live interactive conversational AI demo.
164
20B-parameter Alexa model sets new marks in few-shot learning
With an encoder-decoder architecture — rather than decoder only — the Alexa Teacher Model excels other large language models on few-shot tasks such as summarization and machine translation.
165
I used DALL·E 2 to generate a logo
Everybody has heard about the latest cool thing™, which is DALL·E 2 (henceforth called Dall-e). A few months ago, when the first previews started, it was basically everywhere. Now, a few weeks ago, the floodgates have been opened and lots of people on the waitlist got access - that group included me. I’ve spent a day…
166
167
Differentiable programming from scratch
Differentiable programming has become a hot research topic, and not only due to the popularity of machine learning frameworks like TensorFlow, PyTorch, and JAX. Many fields apart from machine learning are finding differentiable programming to be a useful tool for solving optimization problems. In computer graphics…
168
AlphaFold reveals the structure of the protein universe
Today, in partnership with EMBL’s European Bioinformatics Institute (EMBL-EBI), we’re now releasing predicted structures for nearly all catalogued proteins known to science, which will expand the AlphaFold DB by over 200x - from nearly 1 million structures to over 200 million structures - with the potential to…
169
170
Facial hair tracking for high fidelity performance capture
Facial hair is a largely overlooked topic in facial performance capture. Most production pipelines in the entertainment industry do not have a way to automatically capture facial hair or track the skin underneath it. Thus, actors are asked to shave clean before face capture, which is very often undesirable. In this…
171
172
ML is not that good at predicting consumers' choices
I wanted to share the attached paper on choice prediction that I recently co-authored with Itamar Simonson in case it’s of interest.
173
DALL·E now available in beta
We’ll invite 1 million people from our waitlist over the coming weeks. Users can create with DALL·E using free credits that refill every month, and buy additional credits in 115-generation increments for $15.
174
A 100x Investment (2019)
comma.ai has raised $8.1 million, Zoox has raised $990 million. Cruise and Waymo have raised even more.
175
Thoughts on ML Engineering After a Year of My PhD
My personal website
176
Crimes against transhumanity
(Disclaimer: I am a transhumanist skeptic these days, not to mention a singularity curmudgeon and a critic of Mars colonization, but I still find these ideas nice to chew on sometimes.)
177
ML code generation vs. coding by hand: what we think programming will look like
Will ML code generation like GitHub Copilot replace coding by hand? We say no: developers still maintain generated code, so abstraction matters.
178
30% of Google's Emotions Dataset Is Mislabeled
Last year, Google released their “GoEmotions” dataset: a human-labeled dataset of 58K Reddit comments categorized according to 27 emotions. The problem? A whopping 30% of the dataset is mislabeled! Check out some of the egregious errors, and learn how to build better datasets.30% of Google's Emotions Dataset is…
179
180
Using GPT-3 to explain how code works
One of my favourite uses for the GPT-3 AI language model is generating explanations of how code works. It’s shockingly effective at this: its training set clearly include a vast …