On the Ability of Graph Neural Networks to Model Interactions Between Vertices
ai.papers.bar/paper/530f677c
Graph neural networks (GNNs) are widely used for modeling complex interactions between entities represented as...
๐งต ๐
Follow
Click to Follow labmlai
labml.ai
@labmlai
apps.apple.com/us/app/prompt-
Annotated paper implementations nn.labml.ai
Trending research papers papers.labml.ailabml.aiJoined December 2020
labml.aiโs Tweets
Our team member finetuned an SD 768x768 2.1v model using his wife's photos and experimented with iconic movie characters.
750 training steps, 0.000003 learning rate, and around 20 images used
#stablediffusion #stablediffusionart
๐๐งต
1
5
22
Show this thread
Experimenting with image to image editing with user images
Original image on left taken from our feed: promptart.labml.ai/feed?image_id=
The photo on right is the new generation with our team member
We used #stablediffusion 2.1v model and dreambooth training
๐
1
5
10
Show this thread
Dataset Distillation: A Comprehensive Review
ai.papers.bar/paper/11118b2a
dataset distillation (DD), also known as dataset condensation (DC), has become a popular research topic in recent years. Given an original large...
๐งต ๐
5
29
275
Show this thread
A Statistician Teaches Deep Learning
ai.papers.bar/paper/6c854db5
Deep learning (DL) has gained much attention and become increasingly popular in modern data science. Computer scientists led the way in developing deep...
๐งต ๐
7
114
562
Show this thread
A Survey on Transformers in Reinforcement Learning
ai.papers.bar/paper/b11030ce
Transformer has been considered the dominating neural architecture in NLP and CV, mostly under a supervised setting. Recently, a similar...
๐งต ๐
1
95
426
Show this thread
11/11
Hereโs an annotated PyTorch implementation of stable diffusion for a more detailed look into it :
2
16
Show this thread
Want to train stable diffusion with your own images? Here's what you need to know...
๐งถ๐
1
6
42
Show this thread
Found a bug in the Prompt Art iPhone app's latest version (1.5.6), that crashes whenever you tap on the "create similar" button. This happens for iOS version 16.x
We submitted a build to the app store with a fix and will update once approved.
2
2
Deep Learning่ซๆใฎๅฎ่ฃ
ใพใจใ
github.com/labmlai/annota
ใใฅใผใฉใซใใใใฏใผใฏใจใใใจ้ข้ฃใใใขใซใดใชใบใ ใฎใทใณใใซใชPyTorchๅฎ่ฃ
ใฎใณใฌใฏใทใงใณใๆฏ้ฑๆฐใใๅฎ่ฃ
ใ่ฟฝๅ ใใใฆใใ
143
779
๐ Deep Learning Paper Implementations ๐
A collection of simple PyTorch implementations of neural networks and related algorithms. These implementations are documented with explanations and side-by-side notes.
๐ github.com/labmlai/annota
35
462
1,823
Results for human faces using SD v2.1 and comparison with SD v2.0
You can try it yourself.
iPhone app: apps.apple.com/us/app/prompt-
Web: promptart.labml.ai/playground
#StableDiffusion2
Here are some results left: SDv.1, right: SDv2.1
(1/4)๐งต๐
2
4
13
Show this thread
โจPrompt Art iPhone app version updateโจ
apps.apple.com/app/prompt-art
We added new features including in-painting and image2image with the new version. Thread explains the new additions.
๐๐งต
1
3
13
Show this thread
Our "annotated research paper implementations" Github repo reached 15,000 stars ๐
Github repo: github.com/labmlai/annota
code with side-by-side notes: nn.labml.ai
It has over 50 paper implementations. Thank you everyone for the support!
read image description
ALT
14
75
We reverted to stable diffusion 1.5.
It is possible to pick v2.0 on our web (promptart.labml.ai/playground)
iPhone app uses v1.5 (apps.apple.com/ca/app/prompt-) and we will add the option to chose on the next release
๐๐งต
Quote Tweet
Just updated apps.apple.com/ca/app/prompt- and promptart.labml.ai to use stable diffusion 2.0.
It seems much better at understanding the prompt.
Follow @PromptArtApp for updates
Show this thread
1
7
31
Show this thread
Just updated apps.apple.com/ca/app/prompt- and promptart.labml.ai to use stable diffusion 2.0.
It seems much better at understanding the prompt.
Follow for updates
1
2
5
Show this thread
Just updated apps.apple.com/ca/app/prompt- and promptart.labml.ai to use stable diffusion 2.0. It seems much better at understanding the prompt.
Follow for updates
Quote Tweet
Excited to announce the release of Stable Diffusion 2.0!
Many new features in v2:
โข Base 512x512 and 768x768 models trained from scratch with new OpenCLIP text encoder
โข X4 upscaling text-guided diffusion model
โข New โDepth2Imageโ functionality
Blog: stability.ai/blog/stable-di
Show this thread
3
New Prompt Art iPhone app Released
apps.apple.com/us/app/prompt-
We shipped a bunch of features and performance improvements with this release. Thread explains the new additions.
๐๐งต
1
4
10
Show this thread
(2/4) You can signup and get $1 free credit to try it out. Please reply or DM if you have any feedback or suggestions.
1
2
3
Show this thread
We released an API for Stable Diffusion today. It generates 2 images, 50 steps in 2 seconds and 500 images for $1. It's one of the fastest and cheapest Stable Diffusion APIs out there.
promptart.labml.ai/docs
(1/4) ๐งต๐
3
7
41
Show this thread
๐ We did some performance improvements on promptart.labml.ai. It now generates 2 images in 2.5 secs (50 steps).
You can also set seeds and negative prompts now.
Our iOS app: apps.apple.com/gb/app/prompt-
1
9
32
We also made simple iOS app to generate and browse stable diffusion images
apps.apple.com/gb/app/prompt-
2
Show this thread
promptart.labml.ai can now upscale 512x512 images generated with stable diffusion to 2048x2048
We use SwinIR: Image Restoration Using Swin Transformer (2021) to increase resolution
๐ Paper papers.labml.ai/paper/ea90d584
๐ฅ Official Code github.com/JingyunLiang/S
2
28
124
Show this thread
Generate art for free with stable diffusion: promptart.labml.ai
We updated to SD v1.5 model by . v1.5 has been trained for more than twice the number of steps on images with high aesthetic scores compared to v1.4.
We will soon upgrade the inpainting model too.
read image description
ALT
1
12
39
Show this thread
a collection of simple PyTorch implementations of neural networks and related algorithms:
6
20
We made a list of 2023 submissions. You can click on the papers to see their source code, videos, and comments on social media.
papers.labml.ai/lists/iclr_202
We will update it to show the status when the reviews are out
24
83
โa collection of simple PyTorch implementations of neural networks and related algorithms.โ bit.ly/3g6TAw9
#AI #ML #DataScience #MachineLearning
3
50
163
So exciting to see our work on FlashAttention having an impact, beyond NLP models๐๐๐! Thanks for the integration effort
Quote Tweet
13
118
1
3
18
Show this thread
2/3
FlashAttention: Fast and Memory-Efficient Exact Attention with IO-Awareness
๐ papers.labml.ai/paper/e2031bfa
๐ฅ Github: github.com/HazyResearch/f
They divide Q,K,V into blocks and compute blockwise such that I/O between GPU SRAM to GPU high bandwidth memory is minimized.
๐
1
3
23
Show this thread
Speed up stable diffusion by ~50% using flash attention
๐Annotated implementation: nn.labml.ai/diffusion/stab
๐ฅ Github: github.com/labmlai/annota
We got close to 50% speedup on A6000 by replacing most of cross attention operations in the U-Net with flash attention
๐งถ๐
9
95
499
Show this thread
ๆทฑๅฑคๅญฆ็ฟใฎๆ่ฟใฎ็ ็ฉถ๏ผไธป่ฆใช็ ็ฉถใฎPyTorchๅฎ่ฃ
้ใชใฎใงใใ๏ผใใใฏใใชใใใใงใ
nn.labml.ai
ๅใซgithubใซใณใผใใ่ผใใฆๅฐใใณใกใณใใใฆ็ตใใ๏ผใงใฏใชใ๏ผๅฐ็จใฎใตใคใใง๏ผๆฐๅผใจใณใผใใฎๅฎ่ฃ
้จๅใฎๅฏพๅฟใ่งฃ่ชฌใใฆใใ๏ผใใพใ็ฅใใชใๅ้ใงใใใใใใใ
2
302
1,816
๐ You can now click on maths expressions to highlight them across the page!
Annotated deep learning paper implementations in :
๐ nn.labml.ai
We just added this feature so let us know if you come across any bugs and have thoughts for improvements.
4
65
466
Show this thread
Deep Leaning Paper Implementations - is a good collection of popular deep learning papers implemented in PyTorch.
The codes and documents are rendered side by side which makes it easier for us to understand.
Link: nn.labml.ai
5
87
367
3/3
Stable diffusion is based on "High-Resolution Image Synthesis with Latent Diffusion Models"
๐ papers.labml.ai/paper/2112.107
Latent diffusion models maps from image space to latent space using a pre-trained auto-encoder, and denoises in latent space.
2
1
8
Show this thread
2/2
This implementation based on the official implementation by
github.com/CompVis/stable
We have deployed a server to try stable diffusion here:
promptart.labml.ai
1
3
13
Show this thread
Implementation/tutorial of stable diffusion with side-by-side notes
๐ Code & notes: nn.labml.ai/diffusion/stab
๐ฅ Github: github.com/labmlai/annota
It has annotated model code; DDIM and DDPM sampling; and scripts to generate and in-paint.
๐งถ๐
4
94
411
Show this thread