Updates with Poplar SDK 3.1 release

Author: adamjsanders277

README.md

@@ -10,7 +10,7 @@ To see what's new and easily filter applications by domain and framework, please
 
 For more detailed benchmark information, please visit our [Performance Results page](https://www.graphcore.ai/performance-results).
 
-> The code presented here requires using Poplar SDK 3.0.x
+> The code presented here requires using Poplar SDK 3.1.x, and has been tested using Ubuntu 20.04 and Python 3.8
 
 Please install and enable the Poplar SDK following the instructions in the [Getting Started](https://docs.graphcore.ai/en/latest/getting-started.html#pod-system-getting-started-guides) guide for your IPU system.
 
@@ -39,32 +39,26 @@ If you require POD128/256 setup and configuration for our applications, please c
 
 | Model | Domain | Type |Links |
 | ------- | ------- |------- | ------- |
-| ResNet  | Image Classification | Training & Inference | [TensorFlow 1](vision/cnns/tensorflow1/) , [TensorFlow 2](vision/cnns/tensorflow2/), [PyTorch](vision/cnns/pytorch/), [PyTorch Lightning](https://github.com/graphcore/pytorch-lightning-examples/tree/release/applications)|
-| ResNeXt  | Image Classification | Training & Inference | [TensorFlow 1](vision/cnns/tensorflow1/) , [PopART (Inference)](vision/resnext_inference/popart), [PyTorch (Inference)](vision/cnns/pytorch/inference)
-| EfficientNet | Image Classification | Training & Inference | [TensorFlow 1](vision/cnns/tensorflow1/) , [PyTorch](vision/cnns/pytorch/), [PyTorch Lightning](https://github.com/graphcore/pytorch-lightning-examples/tree/release/applications)|
-| MobileNet | Image Classification | Inference | [TensorFlow 1](vision/cnns/tensorflow1/inference) |
-| MobileNetv2 | Image Classification | Inference | [TensorFlow 1](vision/cnns/tensorflow1/inference) |
+| ResNet  | Image Classification | Training & Inference | [TensorFlow 2](vision/cnns/tensorflow2/), [PyTorch](vision/cnns/pytorch/), [PyTorch Lightning](https://github.com/graphcore/pytorch-lightning-examples/tree/release/applications)|
+| ResNeXt  | Image Classification | Training & Inference | [PopART (Inference)](vision/resnext_inference/popart), [PyTorch (Inference)](vision/cnns/pytorch/inference)
+| EfficientNet | Image Classification | Training & Inference | [PyTorch](vision/cnns/pytorch/), [PyTorch Lightning](https://github.com/graphcore/pytorch-lightning-examples/tree/release/applications)|
 | MobileNetv3 | Image Classification | Training & Inference | [PyTorch](vision/cnns/pytorch/) |
 | ViT(Vision Transformer) | Image Classification | Training| [PyTorch](vision/vit/pytorch/), [Hugging Face Optimum](https://huggingface.co/Graphcore/vit-base-ipu) |
 | DINO | Image Classification | Training| [PyTorch](vision/dino/pytorch) |
 | Swin | Image Classification | Training | [PyTorch](vision/swin/pytorch)  |
 | MAE (Masked AutoEncoder) | Image Classification | Training | [PyTorch](vision/mae/pytorch)  |
-| Yolov3 | Object Detection | Training & Inference | [TensorFlow 1](vision/yolo_v3/tensorflow1) |
 | Yolov4-P5 | Object Detection | Inference | [PyTorch](vision/yolo_v4/pytorch) |
 | Faster RCNN | Object Detection | Training & Inference | [PopART](vision/faster_rcnn/popart) |
 | EfficientDet | Object Detection | Inference | [TensorFlow 2](vision/efficientdet/tensorflow2) |
-| SSD  | Object Detection | Inference | [TensorFlow 1](vision/ssd/tensorflow1)|
 | UNet (Medical) | Image segmentation | Training & Inference | [TensorFlow 2](vision/unet_medical/tensorflow2)  |
-| UNet (Industrial) | Image segmentation | Training | [TensorFlow 1](vision/unet_industrial/tensorflow1)  |
 | Neural Image Fields | Neural Radiance Fields | Training | [TensorFlow 2](vision/neural_image_fields/tensorflow2)  |
 <br>
 
 ### <img width="30" src="https://user-images.githubusercontent.com/81682248/177355208-a49a2bba-dd4d-4467-b135-9023279e8f01.png"></a> Natural Language Processing <a name="nlp"></a>
 
 | Model | Domain | Type |Links |
 | ------- | ------- |------- | ------- |
-| BERT | NLP | Training & Inference |[TensorFlow 1](nlp/bert/tensorflow1) , [PyTorch](nlp/bert/pytorch) , [PopART](nlp/bert/popart), [TensorFlow 2](nlp/bert/tensorflow2/), [PopXL](nlp/bert/popxl), [PaddlePaddle](https://github.com/PaddlePaddle/PaddleNLP/tree/develop/model_zoo/bert/static_ipu), [Hugging Face Optimum](https://huggingface.co/Graphcore/bert-large-ipu)|
-| Group BERT | NLP | Training |[TensorFlow 1](nlp/bert/tensorflow1/README.md#GroupBERT_model) |
+| BERT | NLP | Training & Inference | [PyTorch](nlp/bert/pytorch) , [PopART](nlp/bert/popart), [TensorFlow 2](nlp/bert/tensorflow2/), [PopXL](nlp/bert/popxl), [PaddlePaddle](https://github.com/PaddlePaddle/PaddleNLP/tree/develop/model_zoo/bert/static_ipu), [Hugging Face Optimum](https://huggingface.co/Graphcore/bert-large-ipu)|
 | Packed BERT | NLP | Training |[PyTorch](nlp/bert/pytorch), [PopART](nlp/bert/popart) |
 | GPT2 | NLP | Training |[PyTorch](nlp/gpt2/pytorch) , [Hugging Face Optimum](https://huggingface.co/Graphcore/gpt2-medium-ipu) |
 | GPTJ | NLP | Training |[PopXL](nlp/gpt_j/popxl)|
@@ -86,7 +80,7 @@ If you require POD128/256 setup and configuration for our applications, please c
 | DeepVoice3 | TTS (TextToSpeech) | Training & Inference |[PopART](speech/deepvoice3/popart) |
 | FastSpeech2 | TTS(TextToSpeech) | Training & Inference | [TensorFlow 2](speech/fastspeech2/tensorflow2) |
 | Fastpitch | TTS (TextToSpeech) | Training | [PyTorch](speech/fastpitch/pytorch) |
-| Conformer | STT(SpeechToText) | Training & Inference | [PopART](speech/conformer/popart), [TensorFlow 1](speech/conformer/tensorflow1), [PyTorch](speech/conformer/pytorch)|
+| Conformer | STT(SpeechToText) | Training & Inference | [PopART](speech/conformer/popart), [PyTorch](speech/conformer/pytorch)|
 | Transfomer Transducer | STT(SpeechToText) | Training & Inference | [PopART](speech/transformer_transducer/popart) |
 | Wav2Vec2 | STT(SpeechToText) | Training | [Hugging Face Optimum](https://huggingface.co/Graphcore/wav2vec2-base-ipu)|
 
@@ -108,10 +102,10 @@ If you require POD128/256 setup and configuration for our applications, please c
 
 | Model | Domain | Type |Links |
 | ------- | ------- |------- | ------- |
-| TGN (Temporal Graph Network) | GNN | Training & Inference | [TensorFlow 1](gnn/tgn/tensorflow1) |
 | MPNN (Message Passing Neural Networks) | GNN | Training & Inference | [TensorFlow 2](gnn/message_passing/tensorflow2) |
 | Spektral GNN library with QM9 | GNN | Training | [TensorFlow 2](gnn/spektral/tensorflow2)  |
 | Cluster GCN | GNN | Training & Inference | [TensorFlow 2](gnn/cluster_gcn/tensorflow2) |
+| TGN (Temporal Graph Networks) | GNN | Training | [PyTorch](gnn/tgn/pytorch) |
 
 <br>
 
@@ -120,64 +114,24 @@ If you require POD128/256 setup and configuration for our applications, please c
 | Model | Domain | Type |Links |
 | ------- | ------- |------- | ------- |
 | DeepDriveMD | Biology (Protein folding) | Training | [TensorFlow 2](ai_for_simulation/deep_drive_md/tensorflow2)  |
-| CosmoFlow example using 3D Convolutions  | Cosmology| Training & Inference | [TensorFlow 1](ai_for_simulation/cosmoflow/tensorflow1)|
-| et0 | Evapotransporation  | Inference | [TensorFlow 1](ai_for_simulation/et0/tensorflow1)  |
 | Approximate Bayesian Computation (ABC) COVID-19 | Medical | Inference | [TensorFlow 2](ai_for_simulation/abc_covid_19/tensorflow2)  |
 
-<br>
-
-### <img width="25" src="https://user-images.githubusercontent.com/81682248/177360221-c599b6db-04e7-4e30-8be1-9752085df299.png"></a> Recommender Systems <a name="recommender_systems"></a>
-
-| Model | Domain | Type |Links |
-| ------- | ------- |------- | ------- |
-| Deep AutoEncoders for Collaborative Filtering | Recommender Systems | Training & Inference | [TensorFlow 1](recommendation/autoencoder/tensorflow1) |
-| Click through rate: Deep Interest Network | Recommender Systems | Training & Inference | [TensorFlow 1](recommendation/click_through_rate/tensorflow1) |
-<br>
-
-### <img width="30" src="https://user-images.githubusercontent.com/81682248/177372896-6699b42b-c7ff-4186-93b9-b7f46ec03999.png"></a> Reinforcement Learning <a name="rl"></a>
-
-| Model | Domain | Type |Links |
-| ------- | ------- |------- | ------- |
-| RL Policy model | Reinforcement Learning | Training | [TensorFlow 1](reinforcement_learning/rl_policy_model/tensorflow1) |
 
 <br>
 
 ### <img width="25" src="https://user-images.githubusercontent.com/81682248/177373761-77d40785-5390-400b-ad9f-305f4fd54a05.png"></a> Sparsity <a name="sparsity"></a>
 
 | Model | Domain | Type |Links |
 | ------- | ------- |------- | ------- |
-| MNIST RigL | Dynamic Sparsity | Training | [TensorFlow 1](sparsity/dynamic_sparsity/tensorflow1/mnist_rigl) |
-| Autoregressive Language Modelling | Dynamic Sparsity | Training | [TensorFlow 1](sparsity/dynamic_sparsity/tensorflow1/language_modelling)
-| Block-Sparse library  | Sparsity | Training & Inference | [PopART](sparsity/block_sparse/popart) , [TensorFlow 1](sparsity/block_sparse/tensorflow1)|
-
-
-<br>
-
-### <img width="30" src="https://user-images.githubusercontent.com/81682248/177374313-c567fa25-c1a0-450f-855b-ce8b243d087e.png"></a> Probability <a name="probability"></a>
-
-| Model | Domain | Type |Links |
-| ------- | ------- |------- | ------- |
-| Contrastive Divergence VAE using MCMC methods  | Generative Model | Training | [TensorFlow 1](probability/contrastive_divergence_vae/tensorflow1) |
-| mcmc  | Statistics | Training & Inference | [TensorFlow 1](probability/mcmc/tensorflow1/)|
+| Block-Sparse library  | Sparsity | Training & Inference | [PopART](sparsity/block_sparse/popart)
 
 <br>
 
 ### Miscellaneous <a name="miscellaneous"></a>
 | Model | Domain | Type |Links |
 | ------- | ------- |------- | ------- |
-| Sales forecasting | MLP (Multi-Layer Perceptron) | Training | [TensorFlow 1](miscellaneous/sales_forecasting/tensorflow1) |
 | Monte Carlo Ray Tracing  | Graphics | Inference | [Poplar](miscellaneous/monte_carlo_ray_tracing/poplar) |
 
-<br>
-
-### Archived <a name="archived"></a>
-
-The following applications have been archived. More information can be provided on request.
-| Model | Domain | Type | Framework|
-| ------- | ------- |------- | ------- |
-| Minigo | Reinforcement Learning | Training | TensorFlow 1|
-
-
 <br>
 
 ## Developer Resources
@@ -227,6 +181,17 @@ Unless otherwise specified by a LICENSE file in a subdirectory, the LICENSE refe
 
 ## Changelog
 
+<details>
+<summary>Dec 2022</summary>
+<br>
+
+*  Added this model below to reference models
+    *  GNN: TGN (PyTorch)
+*  Deprecating all PopART applications. Support will be removed in the next release.
+*  Deprecated all TensorFlow 1 applications.
+*  Deprecated Ubuntu 18.04 support.
+</details>
+
 <details>
 <summary>Sep 2022</summary>
 <br>

ai_for_simulation/abc_covid_19/tensorflow2/README.md

@@ -1,78 +1,77 @@
-# Probabilistic COVID-19 modelling with Approximate Bayesian Computation (ABC)
+# Approximate Bayesian Computation (ABC) 
+Approximate Bayesian Computation for probabilistic COVID-19 modelling, optimised for Graphcore's IPU.
 
-## Overview
+| Framework | domain | Model | Datasets | Tasks| Training| Inference | Reference |
+|-------------|-|------|-------|-------|-------|---|---|
+| TensorFlow2 | Simulation | ABC | CSSEGIS COVID-19 | Object detection | ✅ | ✅ | ['Accelerating Simulation-based Inference with Emerging AI Hardware'](https://ieeexplore.ieee.org/document/9325369) |
 
-This is a representative implementation 
-of Approximate Bayesian Computation (ABC)
-for Simulation-based Inference (SBI).
-A detailed explanation is provided in the background section and in
-"Accelerating Simulation-based Inference with Emerging AI Hardware",
-S Kulkarni, A Tsyplikhin, MM Krell, and CA Moritz, 
-IEEE International Conference on Rebooting Computing (ICRC), 2020.
-In its core, the algorithm determines simulation parameters that best
-describe the observed data from COVID-19 infections 
-to enable statistical inference.
 
-## Dataset
+## Instructions summary
 
-The data is provided in the script `covid_data.py`.
-It was extracted from the 
-[JHU CSSE COVID-19 Data](https://github.com/CSSEGISandData/COVID-19).
-It contains the COVID-19 numbers (detected active cases, 
-recovered cases, and detected cases that lead to death) 
-after the first day with 100 known infections.
+1. Install and enable the Poplar SDK (see Poplar SDK setup)
 
-See also: 
-Dong E, Du H, Gardner L. An interactive web-based dashboard
-to track COVID-19 in real time. Lancet Inf Dis. 20(5):533-534.
-doi: 10.1016/S1473-3099(20)30120-1
+2. Install the system and Python requirements (see Environment setup)
 
-## File Structure
+3. Download the CSSEGIS covid-19 dataset (See Dataset setup)
 
-| File                         | Description                                |
-| ---------------------------- | ------------------------------------------ |
-| `README.md`                  | How to run the model                       |
-| `ABC_IPU.py`                 | Main algorithm script to run IPU model     |
-| `argparser.py`               | Document and read command line parameters  |
-| `covid_data.py`              | Provide already downloaded COVID-19 data   |
-| `requirements.txt`           | Required Python 3 packages                 |
-| `test_ABC.py`                | Test script. Run using `python -m pytest`  |
 
-## Quick start guide
+## Poplar SDK setup
+To check if your Poplar SDK has already been enabled, run:
+```bash
+ echo $POPLAR_SDK_ENABLED
+ ```
 
-### 1) Download the Poplar SDK
+If no path is provided, then follow these steps:
+1. Navigate to your Poplar SDK root directory
 
-Install the Poplar SDK following the instructions in the 
-Getting Started guide for your IPU system 
-which can be found here: 
-https://docs.graphcore.ai/projects/ipu-pod-getting-started/en/latest/installation.html.
-Make sure to source the `enable.sh` script for Poplar as well as the drivers.
+2. Enable the Poplar SDK with:
+```bash 
+cd poplar-<OS version>-<SDK version>-<hash>
+. enable.sh
+```
+
+More detailed instructions on setting up your environment are available in the [poplar quick start guide](https://docs.graphcore.ai/projects/graphcloud-poplar-quick-start/en/latest/).
 
-### 2) Package installation
 
-Make sure that the virtualenv package is installed for Python 3.
+## Environment setup
+To prepare your environment, follow these steps:
+
+1. Create and activate a Python3 virtual environment:
+```bash
+python3 -m venv <venv name>
+source <venv path>/bin/activate
+```
 
-### 3) Prepare the TensorFlow environment
+2. Navigate to the Poplar SDK root directory
 
-Activate a Python3 virtual environment with the `tensorflow`
-wheel version 2.4 included in the SDK as follows:
+3. Install the Tensorflow2 and IPU Tensorflow add-ons wheels:
+```bash
+cd <poplar sdk root dir>
+pip3 install tensorflow-2.X.X...<OS_arch>...x86_64.whl
+pip3 install ipu_tensorflow_addons-2.X.X...any.whl
+```
+For the CPU architecture you are running on
 
+4. Build the custom ops:
+```bash
+cd static_ops && make
 ```
-python3 -m venv venv
-source venv/bin/activate
-pip install <path to tensorflow_2.4.whl>
-pip install -r requirements.txt
+
+
+## Dataset setup
+Download the dataset from [the source](https://github.com/CSSEGISandData/COVID-19), or use the script provided:
+```bash
+python3 covid_data.py
 ```
 
-### 4) Execution
 
+## Run the example
 To execute the application, you can run for example
 
-```
-python ABC_IPU.py --enqueue-chunk-size 10000 --tolerance 5e5 --n-samples-target 100 --n-samples-per-batch 400000 --country US --samples-filepath US_5e5_100.txt --replication-factor 4
+```bash
+python3 ABC_IPU.py --enqueue-chunk-size 10000 --tolerance 5e5 --n-samples-target 100 --n-samples-per-batch 400000 --country US --samples-filepath US_5e5_100.txt --replication-factor 4
 ```
 
-All command line options and defaults are explained in `argparser.py`.
 
 ## Background
 We are looking at data from Italy in the Johns Hopkins University dataset
@@ -154,4 +153,4 @@ David J. Warne
 
 - School of Mathematical Sciences, Faculty of Science, Queensland University of Technology, Brisbane, Australia
 - Centre for Data Science, Queensland University of Technology, Brisbane, Australia
-- ARC Centre of Excellence for Mathematical and Statistical Frontiers.
+- ARC Centre of Excellence for Mathematical and Statistical Frontiers.