diff --git a/docs/diffusion/stable_diffusion/latent_diffusion.html b/docs/diffusion/stable_diffusion/latent_diffusion.html
index b74bf0a1..9b71f132 100644
--- a/docs/diffusion/stable_diffusion/latent_diffusion.html
+++ b/docs/diffusion/stable_diffusion/latent_diffusion.html
@@ -76,7 +76,7 @@
#
Latent Diffusion Models
-Latent diffusion models use an auto-encoder to map between image space and latent space. The diffusion model works on the diffusion space, which makes it a lot easier to train. It is based on paper High-Resolution Image Synthesis with Latent Diffusion Models.
+Latent diffusion models use an auto-encoder to map between image space and latent space. The diffusion model works on the diffusion space, which makes it a lot easier to train. It is based on paper High-Resolution Image Synthesis with Latent Diffusion Models.
They use a pre-trained auto-encoder and train the diffusion U-Net on the latent space of the pre-trained auto-encoder.
For a simpler diffusion implementation refer to our DDPM implementation. We use same notations for αt, βt schedules, etc.
diff --git a/docs/index.html b/docs/index.html
index 1f06b6d0..da8d97a6 100644
--- a/docs/index.html
+++ b/docs/index.html
@@ -101,13 +101,15 @@
Vision Transformer (ViT)
Primer EZ
Hourglass
-
-
-
-
-
-
-
+
+
+
+
-
-
+
+
+
+
+
+
+
✨ Graph Neural Networks
-
-Solving games with incomplete information such as poker with CFR.
-
+
+Solving games with incomplete information such as poker with CFR.
+
- Adam
- AMSGrad
@@ -149,15 +156,11 @@
-
+
-
-
Highlighted Research Paper PDFs
diff --git a/docs/papers.json b/docs/papers.json
index 20eb3449..a0f48520 100644
--- a/docs/papers.json
+++ b/docs/papers.json
@@ -123,6 +123,9 @@
"2006.11239": [
"https://nn.labml.ai/diffusion/ddpm/index.html"
],
+ "2010.02502": [
+ "https://nn.labml.ai/diffusion/stable_diffusion/sampler/ddim.html"
+ ],
"2010.07468": [
"https://nn.labml.ai/optimizers/ada_belief.html"
],
@@ -168,6 +171,9 @@
"2112.04426": [
"https://nn.labml.ai/transformers/retro/index.html"
],
+ "2112.10752": [
+ "https://nn.labml.ai/diffusion/stable_diffusion/latent_diffusion.html"
+ ],
"2201.09792": [
"https://nn.labml.ai/conv_mixer/index.html"
],
diff --git a/docs/sitemap.xml b/docs/sitemap.xml
index 5a1dabe4..b9efd7c6 100644
--- a/docs/sitemap.xml
+++ b/docs/sitemap.xml
@@ -134,7 +134,7 @@
https://nn.labml.ai/neox/checkpoint.html
- 2022-08-11T16:30:00+00:00
+ 2022-09-15T16:30:00+00:00
1.00
@@ -533,112 +533,112 @@
https://nn.labml.ai/diffusion/index.html
- 2022-09-12T16:30:00+00:00
+ 2022-09-15T16:30:00+00:00
1.00
https://nn.labml.ai/diffusion/stable_diffusion/util.html
- 2022-09-12T16:30:00+00:00
+ 2022-09-15T16:30:00+00:00
1.00
https://nn.labml.ai/diffusion/stable_diffusion/index.html
- 2022-09-12T16:30:00+00:00
+ 2022-09-15T16:30:00+00:00
1.00
https://nn.labml.ai/diffusion/stable_diffusion/latent_diffusion.html
- 2022-09-12T16:30:00+00:00
+ 2022-09-15T16:30:00+00:00
1.00
https://nn.labml.ai/diffusion/stable_diffusion/sampler/ddim.html
- 2022-09-12T16:30:00+00:00
+ 2022-09-15T16:30:00+00:00
1.00
https://nn.labml.ai/diffusion/stable_diffusion/sampler/index.html
- 2022-09-12T16:30:00+00:00
+ 2022-09-15T16:30:00+00:00
1.00
https://nn.labml.ai/diffusion/stable_diffusion/sampler/ddpm.html
- 2022-09-12T16:30:00+00:00
+ 2022-09-15T16:30:00+00:00
1.00
https://nn.labml.ai/diffusion/stable_diffusion/scripts/text_to_image.html
- 2022-09-12T16:30:00+00:00
+ 2022-09-15T16:30:00+00:00
1.00
https://nn.labml.ai/diffusion/stable_diffusion/scripts/in_paint.html
- 2022-09-12T16:30:00+00:00
+ 2022-09-15T16:30:00+00:00
1.00
https://nn.labml.ai/diffusion/stable_diffusion/scripts/index.html
- 2022-09-12T16:30:00+00:00
+ 2022-09-15T16:30:00+00:00
1.00
https://nn.labml.ai/diffusion/stable_diffusion/scripts/image_to_image.html
- 2022-09-12T16:30:00+00:00
+ 2022-09-15T16:30:00+00:00
1.00
https://nn.labml.ai/diffusion/stable_diffusion/model/unet.html
- 2022-09-12T16:30:00+00:00
+ 2022-09-15T16:30:00+00:00
1.00
https://nn.labml.ai/diffusion/stable_diffusion/model/index.html
- 2022-09-12T16:30:00+00:00
+ 2022-09-15T16:30:00+00:00
1.00
https://nn.labml.ai/diffusion/stable_diffusion/model/clip_embedder.html
- 2022-09-12T16:30:00+00:00
+ 2022-09-15T16:30:00+00:00
1.00
https://nn.labml.ai/diffusion/stable_diffusion/model/autoencoder.html
- 2022-09-12T16:30:00+00:00
+ 2022-09-15T16:30:00+00:00
1.00
https://nn.labml.ai/diffusion/stable_diffusion/model/unet_attention.html
- 2022-09-12T16:30:00+00:00
+ 2022-09-15T16:30:00+00:00
1.00
diff --git a/labml_nn/__init__.py b/labml_nn/__init__.py
index 0ccd4e68..4d961bb4 100644
--- a/labml_nn/__init__.py
+++ b/labml_nn/__init__.py
@@ -41,6 +41,26 @@
* [Primer EZ](transformers/primer_ez/index.html)
* [Hourglass](transformers/hour_glass/index.html)
+#### ✨ [Eleuther GPT-NeoX](neox/index.html)
+* [Generate on a 48GB GPU](neox/samples/generate.html)
+* [Finetune on two 48GB GPUs](neox/samples/finetune.html)
+* [LLM.int8()](neox/utils/llm_int8.html)
+
+#### ✨ [Diffusion models](diffusion/index.html)
+
+* [Denoising Diffusion Probabilistic Models (DDPM)](diffusion/ddpm/index.html)
+* [Denoising Diffusion Implicit Models (DDIM)](diffusion/stable_diffusion/sampler/ddim.html)
+* [Latent Diffusion Models](diffusion/stable_diffusion/latent_diffusion.html)
+* [Stable Diffusion](diffusion/stable_diffusion/index.html)
+
+#### ✨ [Generative Adversarial Networks](gan/index.html)
+* [Original GAN](gan/original/index.html)
+* [GAN with deep convolutional network](gan/dcgan/index.html)
+* [Cycle GAN](gan/cycle_gan/index.html)
+* [Wasserstein GAN](gan/wasserstein/index.html)
+* [Wasserstein GAN with Gradient Penalty](gan/wasserstein/gradient_penalty/index.html)
+* [StyleGAN 2](gan/stylegan/index.html)
+
#### ✨ [Recurrent Highway Networks](recurrent_highway_networks/index.html)
#### ✨ [LSTM](lstm/index.html)
@@ -55,18 +75,6 @@
#### ✨ [U-Net](unet/index.html)
-#### ✨ [Generative Adversarial Networks](gan/index.html)
-* [Original GAN](gan/original/index.html)
-* [GAN with deep convolutional network](gan/dcgan/index.html)
-* [Cycle GAN](gan/cycle_gan/index.html)
-* [Wasserstein GAN](gan/wasserstein/index.html)
-* [Wasserstein GAN with Gradient Penalty](gan/wasserstein/gradient_penalty/index.html)
-* [StyleGAN 2](gan/stylegan/index.html)
-
-#### ✨ [Diffusion models](diffusion/index.html)
-
-* [Denoising Diffusion Probabilistic Models (DDPM)](diffusion/ddpm/index.html)
-
#### ✨ [Sketch RNN](sketch_rnn/index.html)
#### ✨ Graph Neural Networks
@@ -74,12 +82,6 @@
* [Graph Attention Networks (GAT)](graphs/gat/index.html)
* [Graph Attention Networks v2 (GATv2)](graphs/gatv2/index.html)
-#### ✨ [Counterfactual Regret Minimization (CFR)](cfr/index.html)
-
-Solving games with incomplete information such as poker with CFR.
-
-* [Kuhn Poker](cfr/kuhn/index.html)
-
#### ✨ [Reinforcement Learning](rl/index.html)
* [Proximal Policy Optimization](rl/ppo/index.html) with
[Generalized Advantage Estimation](rl/ppo/gae.html)
@@ -88,6 +90,12 @@
[Prioritized Replay](rl/dqn/replay_buffer.html)
and Double Q Network.
+#### ✨ [Counterfactual Regret Minimization (CFR)](cfr/index.html)
+
+Solving games with incomplete information such as poker with CFR.
+
+* [Kuhn Poker](cfr/kuhn/index.html)
+
#### ✨ [Optimizers](optimizers/index.html)
* [Adam](optimizers/adam.html)
* [AMSGrad](optimizers/amsgrad.html)
@@ -119,17 +127,12 @@
* [Fuzzy Tiling Activations](activations/fta/index.html)
-#### ✨ [Sampling Techniques](sampling/index.html)
+#### ✨ [Language Model Sampling Techniques](sampling/index.html)
* [Greedy Sampling](sampling/greedy.html)
* [Temperature Sampling](sampling/temperature.html)
* [Top-k Sampling](sampling/top_k.html)
* [Nucleus Sampling](sampling/nucleus.html)
-#### ✨ [Eleuther GPT-NeoX](neox/index.html)
-* [Generate on a 48GB GPU](neox/samples/generate.html)
-* [Finetune on two 48GB GPUs](neox/samples/finetune.html)
-* [LLM.int8()](neox/utils/llm_int8.html)
-
#### ✨ [Scalable Training/Inference](scaling/index.html)
* [Zero3 memory optimizations](scaling/zero3/index.html)
diff --git a/labml_nn/diffusion/stable_diffusion/latent_diffusion.py b/labml_nn/diffusion/stable_diffusion/latent_diffusion.py
index 1a097e7f..3f08333d 100644
--- a/labml_nn/diffusion/stable_diffusion/latent_diffusion.py
+++ b/labml_nn/diffusion/stable_diffusion/latent_diffusion.py
@@ -12,7 +12,7 @@
latent space. The diffusion model works on the diffusion space, which makes it
a lot easier to train.
It is based on paper
-[High-Resolution Image Synthesis with Latent Diffusion Models](https://arxiv.org/abs/2112.10752).
+[High-Resolution Image Synthesis with Latent Diffusion Models](https://papers.labml.ai/paper/2112.10752).
They use a pre-trained auto-encoder and train the diffusion U-Net on the latent
space of the pre-trained auto-encoder.
diff --git a/readme.md b/readme.md
index 7dcf3160..d8ee600e 100644
--- a/readme.md
+++ b/readme.md
@@ -44,19 +44,17 @@ implementations almost weekly.
* [Primer EZ](https://nn.labml.ai/transformers/primer_ez/index.html)
* [Hourglass](https://nn.labml.ai/transformers/hour_glass/index.html)
-#### ✨ [Recurrent Highway Networks](https://nn.labml.ai/recurrent_highway_networks/index.html)
-
-#### ✨ [LSTM](https://nn.labml.ai/lstm/index.html)
-
-#### ✨ [HyperNetworks - HyperLSTM](https://nn.labml.ai/hypernetworks/hyper_lstm.html)
-
-#### ✨ [ResNet](https://nn.labml.ai/resnet/index.html)
-
-#### ✨ [ConvMixer](https://nn.labml.ai/conv_mixer/index.html)
+#### ✨ [Eleuther GPT-NeoX](https://nn.labml.ai/neox/index.html)
+* [Generate on a 48GB GPU](https://nn.labml.ai/neox/samples/generate.html)
+* [Finetune on two 48GB GPUs](https://nn.labml.ai/neox/samples/finetune.html)
+* [LLM.int8()](https://nn.labml.ai/neox/utils/llm_int8.html)
-#### ✨ [Capsule Networks](https://nn.labml.ai/capsule_networks/index.html)
+#### ✨ [Diffusion models](https://nn.labml.ai/diffusion/index.html)
-#### ✨ [U-Net](https://nn.labml.ai/unet/index.html)
+* [Denoising Diffusion Probabilistic Models (DDPM)](https://nn.labml.ai/diffusion/ddpm/index.html)
+* [Denoising Diffusion Implicit Models (DDIM)](https://nn.labml.ai/diffusion/stable_diffusion/sampler/ddim.html)
+* [Latent Diffusion Models](https://nn.labml.ai/diffusion/stable_diffusion/latent_diffusion.html)
+* [Stable Diffusion](https://nn.labml.ai/diffusion/stable_diffusion/index.html)
#### ✨ [Generative Adversarial Networks](https://nn.labml.ai/gan/index.html)
* [Original GAN](https://nn.labml.ai/gan/original/index.html)
@@ -66,10 +64,19 @@ implementations almost weekly.
* [Wasserstein GAN with Gradient Penalty](https://nn.labml.ai/gan/wasserstein/gradient_penalty/index.html)
* [StyleGAN 2](https://nn.labml.ai/gan/stylegan/index.html)
-#### ✨ [Diffusion models](https://nn.labml.ai/diffusion/index.html)
+#### ✨ [Recurrent Highway Networks](https://nn.labml.ai/recurrent_highway_networks/index.html)
-* [Denoising Diffusion Probabilistic Models (DDPM)](https://nn.labml.ai/diffusion/ddpm/index.html)
+#### ✨ [LSTM](https://nn.labml.ai/lstm/index.html)
+#### ✨ [HyperNetworks - HyperLSTM](https://nn.labml.ai/hypernetworks/hyper_lstm.html)
+
+#### ✨ [ResNet](https://nn.labml.ai/resnet/index.html)
+
+#### ✨ [ConvMixer](https://nn.labml.ai/conv_mixer/index.html)
+
+#### ✨ [Capsule Networks](https://nn.labml.ai/capsule_networks/index.html)
+
+#### ✨ [U-Net](https://nn.labml.ai/unet/index.html)
#### ✨ [Sketch RNN](https://nn.labml.ai/sketch_rnn/index.html)
@@ -123,17 +130,12 @@ Solving games with incomplete information such as poker with CFR.
* [Fuzzy Tiling Activations](https://nn.labml.ai/activations/fta/index.html)
-#### ✨ [Sampling Techniques](https://nn.labml.ai/sampling/index.html)
+#### ✨ [Langauge Model Sampling Techniques](https://nn.labml.ai/sampling/index.html)
* [Greedy Sampling](https://nn.labml.ai/sampling/greedy.html)
* [Temperature Sampling](https://nn.labml.ai/sampling/temperature.html)
* [Top-k Sampling](https://nn.labml.ai/sampling/top_k.html)
* [Nucleus Sampling](https://nn.labml.ai/sampling/nucleus.html)
-#### ✨ [Eleuther GPT-NeoX](https://nn.labml.ai/neox/index.html)
-* [Generate on a 48GB GPU](https://nn.labml.ai/neox/samples/generate.html)
-* [Finetune on two 48GB GPUs](https://nn.labml.ai/neox/samples/finetune.html)
-* [LLM.int8()](https://nn.labml.ai/neox/utils/llm_int8.html)
-
#### ✨ [Scalable Training/Inference](https://nn.labml.ai/scaling/index.html)
* [Zero3 memory optimizations](https://nn.labml.ai/scaling/zero3/index.html)
diff --git a/setup.py b/setup.py
index cabff2b9..8c350568 100644
--- a/setup.py
+++ b/setup.py
@@ -5,7 +5,7 @@
setuptools.setup(
name='labml-nn',
- version='0.4.131',
+ version='0.4.132',
author="Varuna Jayasiri, Nipun Wijerathne",
author_email="vpjayasiri@gmail.com, hnipun@gmail.com",
description="🧑🏫 Implementations/tutorials of deep learning papers with side-by-side notes 📝; including transformers (original, xl, switch, feedback, vit), optimizers (adam, radam, adabelief), gans(dcgan, cyclegan, stylegan2), 🎮 reinforcement learning (ppo, dqn), capsnet, distillation, diffusion, etc. 🧠",