Diego Porres@PDillis

Guatemalan 🇬🇹 physicist, Postdoc @CVC_UAB, researching autonomous driving.

 Barcelona
diegoporres.com
Joined October 2011
1,343 Photos and videos
Fun weekend project: explore the latent space with your hand! Using SD-XL turbo and some pre-defined prompts, this is a proof of concept, but we plan to do so much more. Stay tuned :)
7
866
> 1% of context left > claude code: I'veidentified two more simplifications to the code... > me: fml
7
183
Diego Porres retweeted
Walter Simoncini@iWSimoncini
Oct 15
🪗 Your Transformer Is Secretly Elastic 🧵👇 Transformer families come in a few fixed sizes, so you commit to one for your hardware and latency constraints, even as query loads vary. What if a single pre-trained model could serve every budget? We learn one importance ranking over every MLP neuron and attention head, once, with frozen weights. At inference time, the model can then be pruned to any computational budget by dropping its least important units, without retraining. Across 27 vision and text encoders spanning 22M to 8B parameters, performance degrades smoothly up to 60% sparsity. For example, pruning DINOv3 ViT-H+/16 to 50% sparsity removes 427M parameters, for only a 0.8% drop in ImageNet-1k linear probing. Super fun project in collaboration with Dheeraj Varghese (@dhevarghese), supervised by Gertjan J. Burghouts (@gjburghouts) and Cees Snoek (@cgmsnoek) 🌐 Project page: secretly-elastic.ashita.nl
3
16
39
2,557
predicting future latents/views in autonomous driving makes no sense, you are in an ever-changing environment in contrast to robotics where you hardly move at all (at this stage, i imagine they will realize these things as they want to move between rooms or walk outside)
1
4
95
high-level latents (e.g., concepts) should make more sense imo, as in "I will go fast, dunno what's up there, but I should be more careful"
1
32
turns out it means there is a catastrophic error and no driver works on the server so we have to reinstall everything, fml
Diego Porres@PDillis
Oct 9
CUDA error: unknown error coolio
1
96
CUDA error: unknown error coolio
3
175
"the center of the latent space" the trippiest ending, finding another version of yourself waiting to accompany you; this can only happen in real life if the universe is *infinite* and we go 10^10^128 m away
Odyssey
@odysseyml
Oct 8
Today we're launching Odyssey-3, the most powerful foundation world model yet. It sets a new state of the art on Physics-IQ, and powers robots, trains AIs, and generates interactive experiences. It's really cool. Experience the model today, all for free!
71
"At the edge of a generative cultural precipice" (also the title of my CVPR workshop paper) I like how we can tell the training data has a lot of recordings where the cameraman can be seen, as after all, it is another actor in the scene
Odyssey
@odysseyml
Oct 8
Today we're launching Odyssey-3, the most powerful foundation world model yet. It sets a new state of the art on Physics-IQ, and powers robots, trains AIs, and generates interactive experiences. It's really cool. Experience the model today, all for free!
1
1
82
walked down the precipice and almost made it to the screen, but overshot the camera movement and got sent back to square one
25
prompt: " " led me to a never-ending blanket folding hell, loved it all the actions I could make, besides jumping, were folding those blankets for some reason lol
Odyssey
@odysseyml
Oct 8
Today we're launching Odyssey-3, the most powerful foundation world model yet. It sets a new state of the art on Physics-IQ, and powers robots, trains AIs, and generates interactive experiences. It's really cool. Experience the model today, all for free!
3
1
6
366
I did try to make a run for it, but I was always brought back
35
if world models = data compression, then the next logical step is to do federated world models for avoiding to store petabytes and petabytes of data
1
57
Does mixing WMs from two countries lead to a model from an intermediate (whatever this may mean) country?
28
Diego Porres retweeted
vicc
@viccpoes
Sep 24
btw we’re growing our creative team who are the 🐐s of motion design, creative direction, or filmmaking?
vicc
@viccpoes
Sep 24
realtime srefs?
21
10
101
10,152
Diego Porres retweeted
Andrei Bursuc@abursuc
Sep 17
Word on the computer vision streets: “publish or perish” has now become “publish, promote, or perish”
2
2
375
Diego Porres retweeted
QuiverAI
@QuiverAI
Sep 16
Introducing Arrow 2 Our latest and most advanced models for generating precise, editable vector graphics. Higher quality. Faster outputs. Available now in App and API.
171
836
11,166
2,354,372
Diego Porres retweeted
Douglas Gray
@de3ug
Sep 11
Teaching image generation models about hex codes. Did not realize this was a problem till today. #ECCV2026
1
3
21
2,556
With NumColor you can generate images with precise hex/RGB color control. Trained only on FLUX, it transfers zero-shot to SD3, SD3.5, PixArt-α and PixArt-Σ — 4–9× better color accuracy, 10–30× better color harmony. See you at ECCV! @aatiibutt @j_vazquezcorral @Alvaro_kai93
2
3
6
437
Diego Porres retweeted
Krea
@krea_ai
Sep 3
we've built the world's best agents for creative work. beta starts today, comment/rt to get early access 👇
185
107
518
59,425
Diego Porres retweeted
Joan Rodriguez
@joanrod_ai
Aug 31
app.quiver.ai ✌️

QuiverAI

Create, refine, and share vector graphics with QuiverAI — the platform for AI-assisted SVG workflows.

app.quiver.ai
Neil McArdle
@BetterNeil
Aug 31
What have I just witnessed @QuiverAI!!? My prompt drawn before my eyes, as an SVG 👏🙇‍♂️ Prompt and image in 🧵👇
1
1
21
1,830
Load more