Zippy
@AlexanderRedde3
Followers
572
Following
443
Media
36
Statuses
100
Software engineer working on ML image generation techniques & inference. Obsessed with programming and life.
Here and there
Joined March 2022
Made an fp8 implementation of Flux which gets ~3.5 it/s 1024x1024 on 4090, >2x faster than other methods. https://t.co/H3DkO0sZYK
#flux1
github.com
Flux diffusion model implementation using quantized fp8 matmul & remaining layers use faster half precision accumulate, which is ~2x faster on consumer devices. - aredden/flux-fp8-api
0
3
18
For anyone who may think that this is a fast-forwarded video, or using image caching, I can guarantee it is not- the images are all base64 and it's using a websocket, the terminal below is the output from the tensorrt rest-api server in real time as I ctrl+y / ctrl+z
3
0
25
Finally figured out how to speed up my #sdxlturbo frontend! It's so fast that the only way to show the actual speed is to delete the prompt, since I can't type fast enough 😆 .. built with next.js frontend & tensorrt backend.
16
28
221
Now I'm at 38 fps, from switching to taesd- but have no way to view the images that fast 🥹
2
0
6
Optimized further meaning I'm getting 17 fps for full clip encode -> unet pass -> vae decode -> serialize to jpeg bytes.
0
0
5
Now we're real time! 😎 - with SDXL turbo & tensorrt optimized further.
7
2
21
Made a neat GUI in react for fun & hooked it up to an optimized sdxl-turbo tensorrt api backend I built for image autocomplete! It generates so fast that my browser cant keep up 🥲 - so fun! #sdxlturbo #sdxl #AIart #stablediffusion
3
5
17
SDXL unet finetuning is going pretty well! FIgured out how to train without deepspeed on 4090s 😀 #SDXL09 #diffusers #AIArtwork
2
0
10
Blockadelabs built a neat thing! 🥳 (which I built the ML backend for) 😳
REEEMIXXXXX! 📢📢📢 Infinite 360° worlds just got easier to fine tune. Transform anything - including visual style, time of day, & weather while keeping the structure intact! LIVE NOW: https://t.co/degS4wvbpf
#skyboxlab #texttoskybox #aiart #gamedev #aitools
4
3
31
We release a new ViT-G/14 CLIP model with OpenCLIP which achieves 80.1% zero-shot accuracy on ImageNet and 74.9% zero-shot image retrieval (Recall@5) on MS COCO. As of January 2023, this is the best open source CLIP model. https://t.co/TmVTUP3tBx
https://t.co/PMnpUUTNpc
16
165
820
Here is a closeup of the right end, so you can get a sense of detail
0
0
5
Will probably be hard to appreciate via twitter, but- here is a true seamless 15616 by 1024 image 100% generated by stable diffusion 🤯
12
9
69
0
0
12
Video to video with gradient added wrt each real video frame vs generated frame. 😊 This is one of my test results. Various celebs smiling -> Jennifer Lawrence smiling. #stablediffusion #AIart
Showing off our next version of AI text-to-video capabilities featuring Jennifer Lawrence. Check out how smooth is is!
3
7
46