CLIP-guided collage generation - optimizing the positions/sizes/rotations of small objects on a canvas using CLIP and ES. text prompt: "An illustration of a happy face of a boy" Google Colab: colab.research.google.com/drive/1H_g60Q_…

Check out the technology preview for GitHub code search, the next iteration for search, discovery, and navigation on GitHub. github.blog/2021-12-08-imp…

.Gradio demo for ArcaneGAN now on Hugging Face Spaces demo: huggingface.co/spaces/akhaliq… github: github.com/Sxela/ArcaneGAN

I'm the one who trains Original: Daniel Bibart

Seeing your ‘new’ idea in an old paper

So Google AI trained a Transformer for almost 60 days on 1024 TPUv3 chips. It can now count how many apples there are left

Someone's trying to scam my students into buying gift cards for "me". Unfortunately, they're too smart.

I hope R2 can be less obsessed with *novelty* and more with *usefulness*. Most (if not all) papers are “just some combination of known ideas/techniques”.

Do we all agree that whiteboard coding interviews simply test for (high IQ + high motivation + ability to work evenings/weekends) and not actual coding ability? If so, why pretend? Just tell candidates to improve their chess Elo rating by X% and hire them on the spot.

I added the PRK and PLMS sampling methods (from openreview.net/forum?id=PlKWV…) to github.com/crowsonkb/v-di… and the cc12m_1_cfg Colab at colab.research.google.com/drive/1TBo4saF…. The default for the Colab is now 50 steps PLMS which seems to give pretty good quality. Have fun! :)

Cool idea randomly on Twitter deserves more views. 1) Train both a conditioned and unconditioned autoregressive model by random masking, then 2) at test time you get a knob to interpolate (or better, extrapolate!) the conditioning strength, e.g. to "super-condition" a prediction.

I forgot how cool European cities are. More compact, denser, more unique / interesting, cleaner, safer, pedestrian/bike friendly, a lot more pedestrian only plazas with people relaxing / hanging out. A lot more of outside is an outdoor living space, not just transportation space.

Created with DALL·E 2 by OpenAI 📝 "A scientist comparing apples and oranges, by Norman Rockwell." 🔎 These research objects seem to have definitely distinguishing features... #DALLE // #dalle2 // #DALLEmerz

everyone on ML twitter right now

List of large language models and APIs that let people use them. reddit.com/r/GPT3/comment…

Oracle Guided Image Synthesis with Relative Queries abs: arxiv.org/abs/2204.14189 github: github.com/helblazer811/o…

Photorealistic Text-to-Image Diffusion Models with Deep Language Understanding project page: gweb-research-imagen.appspot.com sota FID(7.27 on COCO), without ever training on COCO, human raters find Imagen samples to be on par with the COCO data itself in image-text alignment

AI Generative fill with memes

i love you all. today was a weird experience in many ways. but one unexpected one is that it has been sorta like reading your own eulogy while you’re still alive. the outpouring of love is awesome. one takeaway: go tell your friends how great you think they are.

This Claude 2.1 vs. GPT-4-Turbo chart absolutely blows my mind. Insane levels of fidelity from OpenAI. Credit: Greg Kamradt