Linqi (Alex) Zhou
@linqi_zhou
Followers
568
Following
240
Media
7
Statuses
56
Research Scientist @LumaLabsAI. Ph.D. Student at Stanford University (on leave). Prev co-founder @apparatelabs (acq.).
Los Angeles, CA
Joined August 2019
📢 Diffusion models (DM) generate samples from noise distribution, but for tasks such as image-to-image translation, one side is no longer noise. We present Denoising Diffusion Bridge Models, a simple and scalable extension to DMs suitable for distribution translation problems.
4
46
257
📢Text-to-3D generation via score distillation (DreamFusion, ProlificDreamer, etc.) produces high-quality 3D assets, but can take up to 10 hours to run. We present an acceleration method for all existing approaches based on score distillation and achieves up to 4.7x speedup.
3
25
138
Amazing team and amazing product. Glad to have been a part of it.
Introducing Pika 1.0, the idea-to-video platform that brings your creativity to life. Create and edit your videos with AI. Rolling out to new users on web and discord, starting today. Sign up at
8
6
108
Check out our work 3D Shape Generation and Completion through Point-Voxel Diffusion (PVD)! PVD probabilistically and diversely completes partial shapes from (real) depth scans and achieves SOTA on shape generation. w/ @du_yilun and @jiajunwu_cs.Website:
1
5
49
Our Huggingface page can be found here:
📢 Diffusion models (DM) generate samples from noise distribution, but for tasks such as image-to-image translation, one side is no longer noise. We present Denoising Diffusion Bridge Models, a simple and scalable extension to DMs suitable for distribution translation problems.
1
3
37
1
0
27
Our model can theoretical subsume many prior frameworks. We additionally propose improved network parameterization and hybrid SDE+ODE sampler for SOTA performance. w/ @aaron_lou @samar_a_khanna @StefanoErmon . Paper url: Code:
0
2
15
Paper: Huggingface: Project page: Code: Shout out to @andyshih_ @chenlin_meng @StefanoErmon for the amazing collaboration!.
2
2
13
Fortunate to have contributed to this amazing work.
Diffusion Model Alignment Using Direct Preference Optimization . abs: This paper from Stanford and Salesforce reformulates DPO in a tractable manner for diffusion models, applied to SDXL with Pick-a-Pic dataset to obtain DPO-SDXL, which outperforms
0
0
10
We extend diffusion processes to arrive at given endpoints, and propose “denoising bridge score matching“ to learn the intractable score for reversing this bridge process. We can exactly sample noisy views given data. The resulting algorithm is simple and scalable.
1
1
10
@andyshih_ @chenlin_meng @StefanoErmon Special thanks to @chenlin_meng and @pika_labs for their support as this is also a work done at Pika.
0
0
5
The website has experienced a short downtime and we lost the images. The images are now back. Apologies to those who viewed during this time.
0
0
1
@samuel_ys92 Our framework focuses more on the development of the theory and is orthogonal to ControlNet, which is an architecture fine-tuning pretrained diffusion. Our framework can also use ControlNet architecture and fine-tune from pretrained diffusion weights.
0
0
1
@gdb This is general purpose in what sense? I still see that it specializes in solving Rubik’s cube no?.
0
0
1