darthgera123 (@PulkitRega) header
darthgera123 (@PulkitRega) profile picture

PhD @cvssp_research |Applied Scientist @Flawlessai | Computer Vision | Ex: @VcaiMpi @iiit_hyderabad| Ryan Gosling Meme Enjoyer

London, UK
Joined June 2014
Wendell Pierce plays the same character in every show and keeps killing it
361
Messaging ICLR reviewers like a crazy ex and getting ghosted. Why even bother with 2 week review phase #ICLR2027
2
25
3,835
If reviewers are not even going to bother what Claude gives as slop and ask for pointless experiments which claude loves suggesting. Can we then stop pretending and just use claude as a judge. Getting 3 identical reviews is insane for ICLR #iclr
3
3
3
86
6,708
Underrated show all breaking bad enjoyers should watch: Fargo. So similar in style and insanely stacked cast
3
199
If people are looking for some really premier TV show.... No look further than The Wire. Its extremely slow. It will take time for you to make sense of it. But its a show that stays with you. Very few shows can do that.
1
3
107
Can someone RL finetune Claude to not be so so so so so verbose. I get a headache reading it. Also when i say "Lets build a pipeline and then we will implement" why is it already implementing !!!!!
4
255
darthgera123 retweeted
Say hello to MOSS-VL-Realtime! This 11B, 256K-context VLM brings video understanding into the stream. 🚀 License: Apache 2.0. 🤖 modelscope.ai/models/openmos… 🏆 Streaming benchmarks: 70.2 avg on OVOBench, 47.2 avg on ProactiveVideoQA, and 66.0 on OmniMMI Proactive Alerting, ahead of key open-source streaming baselines. 🎥 Realtime + interruptible: process timestamped frames continuously and answer questions at any moment based on frames observed so far. 🤫 Proactive silence + correction: stays silent when visual evidence is insufficient, then updates responses as the scene changes. ⚙️ App-ready: session/queue-style online inference, offline image/video APIs, plus Instruct and Base checkpoints for broader use.
8
52
4,222
Happy to share our #CVPR2026 Oral "Relightable Holoported Characters"! Capture a person once, relight them in any environment — from any viewpoint ✨ #ComputerVision #ComputerGraphics
2
7
1
43
5,473
Is it just me or its heartbreaking to see people working on their mac's while in beautiful locations.... Please please look beyond the laptop
224
I have a whimsical question. Is coding or computer use the benchmark for human level intelligence? Because that would be quite sad
Dario Amodei says we're close to human-level AI, yet society doesn't see the tsunami coming Instead of preparing, people dismiss it as a trick of the light, while some push to accelerate blindly without realizing the risks "technical control has improved more than expected, but societal awareness has gotten worse"
178
Great blog!! VGGT was 1 paper that changed my outlook on how Vision is evolving. However i feel the debate here is driven by interfaces. Robotics is one interface which biases this debate. AR/VR is another interface & it will argue about 3D being better.
In my recent blog post, I argue that "vision" is only well-defined as part of perception-action loops, and that the conventional view of computer vision - mapping imagery to intermediate representations (3D, flow, segmentation...) is about to go away. vincentsitzmann.com/blog/bit…
4
304
I miss summer :(
2
129
This paper could only come out of @iiit_hyderabad xD. Proud Alumni. Its missing my hostel mess biryani tho xD
my favourite citation yet
188
Check our paper out at #ICCV2025 and define new benchmarks for real fullbody relighting tasks
Glad to have our #ICCV2025 work out in open. It was a very rewarding experience learning and setting up the lightstage, capturing data and then creating Gaussian Avatars. First of its kind for developing relighting human avatars
3
447
Phenomenal work
We present UniGAHA: Audio-Driven Universal Gaussian Head Avatars at #SIGGRAPHAsia2025 From speech, UniGAHA generates photoreal 3D head avatars that lip-sync, shift gaze, and capture subtle expressions. Works across identities + adapts from a short video. bit.ly/46z7p0G
2
279
Excited to see this work out in the open. Amazing project
🧵 PractiLight: Practical Light Control Using Foundational Diffusion Models 🔗 yoterel.github.io/PractiLigh… Diffusion models already “know” a lot about light transport. No need for massive finetuning that hurts generalization. {1/6} 👇
1
1
3
503
Glad to have our #ICCV2025 work out in open. It was a very rewarding experience learning and setting up the lightstage, capturing data and then creating Gaussian Avatars. First of its kind for developing relighting human avatars
Breaking news: A large-scale, publicly accessible dataset of multi-view OLAT full-body human captures now exists. Accepted at #ICCV2025! A collaboration between MPI-INF and NVIDIA led by Timo Teufel. For @jankautz and me, this is our second joint work. vcai.mpi-inf.mpg.de/projects…
4
2
1
26
2,781
darthgera123 retweeted
Developed a more generalized version. The pattern 東南西北 can be displaced, but remains conserved.
37
239
27
3,078
240,374
Interesting idea. I remember Psycho Pass also had a similar idea of detecting future crime by analyzing the worst criminals in the society
Google just released a brilliant research, a new active learning method for curating high-quality data that reduces training data requirements for fine-tuning LLMs by orders of magnitude. You get the same or better model quality with 250 to 450 expert labels instead of 100K, thanks to focusing expert effort only on confusing boundary cases. That means dramatically lower labeling cost, much faster iteration, up to 10,000x less data to collect, and easier updates when policies or abuse patterns change. Targeted expert labeling around the decision boundary lets a small fine‑tuning set of 250 to 450 examples match or beat models trained on 100K crowdsourced labels, reaching up to 10,000x training data reduction without hurting quality. The usual practice, throwing massive noisy datasets at the model, is slow, expensive, and brittle when ad safety rules or abuse patterns shift. 🧠 The Core idea The team uses an LLM as a scout to sweep a huge pool of ads, then asks experts to judge only the handful of examples that truly confuse the model. Those expert decisions train the next model iteration, and the loop repeats until model–expert agreement stops improving. It is classic active learning, but adapted to large, noisy, imbalanced traffic like ads, where only about 1% of items are actually problematic. @GoogleResearch
143