Sign in

Christian Laforte

@chrlaf.bsky.social
1.4K followers 1.5K following 263 posts

ML Engineer at NVIDIA. Previously: Stealth GPU startup; Stability AI; AMD; Autodesk; CEO of 2 startups (3D + AI). Toronto, Canada

PostsRepliesMedia
Christian Laforte @chrlaf.bsky.social · 10/05/2026
With your skill set, maybe try something remarkable and funny, like Joseph Redmond’s “my little resume”
A resume for a computer scientist, that looks like an advertisement for “my little pony”
220
Christian Laforte @chrlaf.bsky.social · 01/05/2026
NVIDIA researchers published "MotionBricks", real-time 3D character animation for games and robots: Project: nvlabs.github.io/motionbricks/ Check out that demo video... very impressive!
1218
Christian Laforte @chrlaf.bsky.social · 26/02/2026
This is the actual reason he’s obsessed with mars, rockets, and nazis: he was named for an emperor of mars in a nazi sci-fi novel.
Google search for “elon musk named after nazi scifi”.
001
Christian Laforte @chrlaf.bsky.social · 24/02/2026
I'm late to the game but... making crazy combos in @playbalatro.com is awesome!
The "You Win!" screen with my final score, that exceeded 25 millions in one hand.
010
Christian Laforte @chrlaf.bsky.social · 22/05/2025
Here are my young pawpaws, when they bloomed a few days ago! We live a bit north of Toronto. My sister who lives near Sherbrooke, Quebec, is also trying to grow some but that’s beyond the northern limit… we’ll see how that works out!
Young pawpaw tree that bloomed in early MayMe standing behind one of my young pawpaw trees, wearing an old red AMD polo shirt. The tree is about as high as my chin, still growing fast.
020
Christian Laforte @chrlaf.bsky.social · 01/03/2025
I would happily pay $10/month for an automatic algorithm like Twitter's so I can completely disconnect from X. Ideally with a few options, e.g. turn on/off politics temporarily.
110
Christian Laforte @chrlaf.bsky.social · 01/03/2025
- their forward diffusion process consists in masking tokens independently - they train the model to gradually unmask tokens at arbitrary locations in the sequence I find it counter-intuitive that this would work at all...
210
Christian Laforte @chrlaf.bsky.social · 01/03/2025
Project: ml-gsai.github.io/LLaDA-demo/ @apolinario.bsky.social mentioned this demo: huggingface.co/spaces/multi... (video from Apolinario. It's great to understand how the token sequence is generated!)
110
Christian Laforte @chrlaf.bsky.social · 01/03/2025
LLaDA - Large Language Diffusion Models arxiv.org/abs/2502.09992 Significant progress towards language diffusion models. Reportedly on par with LLaMA3 on many benchmarks.
Graph of 6 benchmarks where generally, LLaDA 8B base beats LLaMA2 7B base, and is on par with LLaMA3 8B Base
290
Christian Laforte @chrlaf.bsky.social · 28/02/2025
I've long admired NVIDIA from afar. Now I'm excited to see what makes them successful from the inside; to meet new colleagues; and to make my own contribution! #NVIDIAlife
an NVIDIA banner in white and green. "I can't wait to see what's next" is written on it.
150
Christian Laforte @chrlaf.bsky.social · 28/02/2025
Unsurprisingly, they concluded that flow matching diffusion models (as of now) result in better quality. It's a very long paper, but reasonably accessible to newcomers... if you're interested in that space, I recommend you start with reading their conclusion and discussion on page 59:
110
Christian Laforte @chrlaf.bsky.social · 28/02/2025
Despite having never been trained on unrealistic videos or anime, it can work surprisingly well on them: comfyanonymous.github.io/ComfyUI_exam... comfyanonymous.github.io/ComfyUI_exam...
100
Christian Laforte @chrlaf.bsky.social · 24/11/2024
Another comparison. Very smooth, highly detailed reconstruction:
Fig. 7. Side-by-side comparison. We compare the final inverse rendering of SDF Convolution (8 auxiliary spp), SDF Reparameterization, and our method
using the same optimization setup: in total we run 5000 iterations optimizing 50 views; in each iteration, we optimize a batch of 5 views, rendered with
512 × 512 resolution and 64 spp. In all test cases, our method results in comparable or more accurate reconstructions.
100
Christian Laforte @chrlaf.bsky.social · 24/11/2024
They are able to reconstruct fine details cleanly:
[impressive 3D reconstruction of a plan with thin branches and leaves, with fine details]
Figure official description:
The jump discontinuities at the visibility boundary B make physically based rendering non-differentiable. To explicitly sample and address such discontinuities, we propose to relax the visibility boundary to a thin band of relaxed boundary A. This corresponds to relaxing the surfaces to thin volumes for sampling discontinuities but retaining surface representations for forward rendering. Consequently, we are able to compute accurate gradients of the rendered images with respect to the scene parameters and, in turn, achieve high-quality inverse rendering results, with less complexity than previous methods.
120
Christian Laforte @chrlaf.bsky.social · 24/11/2024
Impressive results: "ViewExtrapolator effectively refines the artifact-prone renderings (left side of arrows) of radiance fields or point clouds, to more realistic renderings with fewer artifacts (right side of arrows)"
Figure 1 shows inputs from point clouds with lots of holes, and outputs where the holes were filled with plausible 3D reconstruction
100
Christian Laforte @chrlaf.bsky.social · 23/11/2024
Figure 2. Overview of SAMURAI visual object tracker
100
Christian Laforte @chrlaf.bsky.social · 23/11/2024
Overview pipeline of SAMPart3D:
Figure 2. An overview pipeline of SAMPart3D. (a) We first pre-train 3D backbone PTv3-object on 3D large-scale data Objaverse, distilling
visual features from FeatUp-DINOv2. (b) Next, we train light-weight MLPs to distill 2D masks to scale-conditioned grouping. (c) Finally,
we cluster the feature of point clouds and highlight the consistent 2D part area with 2D-3D mapping on multi-view renderings, and then
query semantics from MLLMs
000
Christian Laforte @chrlaf.bsky.social · 19/11/2024
Looks like Bluesky just passed 20M users! bcounter.nat.vg congrats to that team for the nearly-flawless scaling and the moderation still holding up
Over 20M users according to the counter.nat.vg web site
070