Jack (in SF) Langerman @jacklangerman.bsky.social · 03/06/2026🎉Workshop On Urban Scene at @cvprconference.bsky.social modeling kicks off in less than 1 hr!!🎉 first up: @niessner.bsky.social ! You don't want to miss this one!! Mile high 3B 010
Jack (in SF) Langerman @jacklangerman.bsky.social · 03/06/2026We have arrived @cvprconference.bsky.social ! it's going to be a great week! 010
Jack (in SF) Langerman @jacklangerman.bsky.social · 03/06/2026🌟🌟 Start @CVPR off right with the Workshop on Urban Scene Modeling! (Tues Jun 2) @ Mile High 3B! 🌟 - Great lineup of key note speakers! - Challenge winners will be announced! - Selected paper talks! 3D/4D reconstruction+generation+understanding applied to the built world 100
Jack (in SF) Langerman @jacklangerman.bsky.social · 22/09/2025what makes a good paper video? what are some of the best you have seen? cc: @davidpicard.bsky.social @davidbau.bsky.social @eugenevinitsky.bsky.social @chriswolfvision.bsky.social @nsaphra.bsky.social @ducha-aiki.bsky.social @parskatt.bsky.social @martin-r-oswald.bsky.social @ericzzj.bsky.social 221
Jack (in SF) Langerman @jacklangerman.bsky.social · 30/04/2025hey @cloneofsimo.bsky.social (or anyone else) did you ever try LoRA+ style LR split between down/up LoRA projection matrices on diffusion models? 010
Jack (in SF) Langerman @jacklangerman.bsky.social · 17/12/2024lol 4o just decided to throw this rocket emoji into a normal technical conversation for no apparent reason who's overly excited discord / slack is responsible for this? 120
Jack (in SF) Langerman @jacklangerman.bsky.social · 14/12/2024"I'm a programmer, not a reptile. I have a Python logo on my gear, but I don't have scales or a slithery body." x.com/voooooogel/s... 110
Jack (in SF) Langerman @jacklangerman.bsky.social · 14/12/2024Anyone have opinions on Mac Book Air M3 as a personal laptop if I still want to mess with some models on the weekend now and then? 110
Jack (in SF) Langerman @jacklangerman.bsky.social · 11/12/2024Locating and Editing Factual Associations in GPT arxiv.org/abs/2202.05262 1. to discover location of facts: at decisive token (eg Michael Jordan plays <BLANK>) iteratively remove circuits from model until <BLANK> != "basketball" 2. edit facts via Rank-1 modifications (ROME) 011
Jack (in SF) Langerman @jacklangerman.bsky.social · 11/12/2024Plug-and-Play Diffusion Features for Text-Driven Image-to-Image Translation pnp-diffusion.github.io arxiv.org/abs/2211.12572 run two copies of model A: DDIM inversion of source image B: generation with target prompt inject feature/attention maps from A into B during generation 111
Jack (in SF) Langerman @jacklangerman.bsky.social · 11/12/2024Steering Language Models With Activation Engineering arxiv.org/abs/2308.10248 ActAdd manipulates the residual stream using something like classifier free guidance: it adds the difference vector between activations for prompts with and without the target characteristic to the residual stream 111
Jack (in SF) Langerman @jacklangerman.bsky.social · 07/12/2024I know the ROME/Memit stuff from @davidbau.bsky.social's lab and this automated circuit discovery stuff. Any other model editing/applied mech interp things people think are particularly practical? Have you used them in practice? 210
Jack (in SF) Langerman @jacklangerman.bsky.social · 05/12/2024I have arrived in SF, and so far it is ....... exciting 240