Completed
Per Chenyang Yuan: at 10:15, the blurry image that results when removing random noise in DDPM is probably due to a mismatch in noise levels when calling the denoiser. When the denoiser is called on x…
Class Central Classrooms beta
YouTube videos curated by Class Central.
Classroom Contents
How AI Videos Actually Work - Diffusion Models, CLIP, and the Math of Turning Text into Images
Automatically move to the next video in the Classroom when playback concludes
- 1 0:00 - Intro
- 2 3:37 - CLIP
- 3 6:25 - Shared Embedding Space
- 4 8:16 - Diffusion Models & DDPM
- 5 11:44 - Learning Vector Fields
- 6 22:00 - DDIM
- 7 25:25 Dall E 2
- 8 26:37 - Conditioning
- 9 30:02 - Guidance
- 10 33:39 - Negative Prompts
- 11 34:27 - Outro
- 12 35:32 - About guest videos + Grant’s Reaction
- 13 6:15 CLIP: Although directly minimizing cosine similarity would push our vectors 180 degrees apart on a single batch, overall in practice, we need CLIP to maximize the uniformity of concepts over the…
- 14 Per Chenyang Yuan: at 10:15, the blurry image that results when removing random noise in DDPM is probably due to a mismatch in noise levels when calling the denoiser. When the denoiser is called on x…
- 15 For the vectors at 31:40 - Some implementations use fx, t, cat + alphafx, t, cat - fx, t, and some that do fx, t + alphafx, t, cat - fx, t, where an alpha value of 1 corresponds to no guidance. I cho…
- 16 At 30:30, the unconditional t=1 vector field looks a bit different from what it did at the 17:15 mark. This is the result of different models trained for different parts of the video, and likely a re…
