Added samples: GitHubLabeler and GettingStarted #198

OliaG · 2018-05-22T05:51:57Z

Iris sample
Sentiment Analysis sample
TaxiFare sample
GitHub issues classification sample

- Iris sample - Sentiment Analysis sample - TaxiFare sample -GitHub issues classification sample

asthana86

👍

Would recommend Adding a few comments around training and evaluation pipelines for the iris, taxi and github samples as well similar to the sentiment analysis one.

justinormont · 2018-05-22T07:13:38Z

Samples/GettingStarted/BinaryClassification_SentimentAnalysis/Program.cs

+
+            //add a FastTreeBinaryClassifier, the decision tree learner for this project, and 
+            //three hyperparameters to be used for tuning decision tree performance 
+            pipeline.Add(new FastTreeBinaryClassifier() {NumLeaves = 5, NumTrees = 5, MinDocumentsInLeafs = 2});


I'd recommend Bigrams+Trichargrams w/ AveragedPerceptronBinaryClassifier{iter=10} for text. AveragedPerceptron generally wins on text vs. FastTree in terms of accuracy & speed.

Thank you! I will use them for advance scenario samples. This one is a "Hello world" type of sample where I'm trying to make the pipeline as simple as possible. And in the next samples I'll demonstrate how the results can be improved with different transforms.

justinormont · 2018-05-22T07:16:59Z

Samples/GettingStarted/BinaryClassification_SentimentAnalysis/Program.cs

+
+            // TextFeaturizer is a transform that will be used to featurize an input column. 
+            // This is used to format and clean the data.
+            pipeline.Add(new TextFeaturizer("Features", "SentimentText"));


Would be a good opportunity to show off some of the hyperparameters of the TextFeaturizer.

pipeline.Add(new TextFeaturizer("Features", "SentimentText") { WordFeatureExtractor = new NGramNgramExtractor() { NgramLength = 2, AllLengths = true }, CharFeatureExtractor = new NGramNgramExtractor() { NgramLength = 3, AllLengths = false } });

Others of course are available:

machinelearning/test/Microsoft.ML.Tests/Scenarios/SentimentPredictionTests.cs

Line 54 in 86f5ee6

pipeline.Add(new TextFeaturizer("Features", "SentimentText")

This makes sense in other samples we are adding or going to add. For getting started sample the easier the better.

justinormont · 2018-05-22T07:22:38Z

Samples/GettingStarted/MulticlassClassification_Iris/Program.cs

+            Console.WriteLine($"    AccuracyMacro = {metrics.AccuracyMacro:0.####}, a value between 0 and 1, the closer to 1, the better");
+            Console.WriteLine($"    AccuracyMicro = {metrics.AccuracyMicro:0.####}, a value between 0 and 1, the closer to 1, the better");
+            Console.WriteLine($"    LogLoss = {metrics.LogLoss:0.####}, the closer to 0, the better");
+            Console.WriteLine($"    LogLoss for class 1 = {metrics.PerClassLogLoss[0]:0.####}, the closer to 0, the better");


Would it be possible to get the actual name for class 1? When a user runs on their own dataset, they will likely be interested in their actual class names.

shauheen · 2018-05-22T14:06:41Z

This PR closes #140 @OliaG

markusweimer

I did not review the samples in detail. However, I suggest some overall changes:

It would be good to have XML Docs on the major classes and methods in each project.
Where will the long form description of each sample go? My vote would be for a README.md in each of the project folders.

markusweimer · 2018-05-22T14:10:18Z

Samples/GitHubLabeler/GitHubLabeler/App.config

+<?xml version="1.0" encoding="utf-8" ?>
+<configuration>
+  <appSettings>
+    <!--TODO: Please enter your own credentials here. -->


We should add a warning to this such that people know that they should not commit this file to a public repo, once changed.

How far would adding this to the .gitignore go for helping folks to not commit this back to a repo? I've done regex checking in githooks before (eg: /(password|access_token) ?=/i), but I don't think githooks can be added to a repo & auto-run by other users.

I don't think there is a great solution for this. In general, the recommendation is to avoid using credentials from code and instead use certificates, user-based authentication, or a service like Azure KeyVault. Unfortunately this doesn't work for samples very well...

TomFinley · 2018-05-22T14:47:44Z

Samples/GitHubLabeler/.gitignore

@@ -0,0 +1,288 @@
+## Ignore Visual Studio temporary files, build results, and
+## files generated by popular Visual Studio add-ons.
+##


I don't think checking in this .gitignore file was intentional, was it?

Agreed. We should only use the one from the root.

TomFinley · 2018-05-22T14:50:58Z

Samples/GettingStarted/Regression_TaxiFarePrediction/Program.cs

+    {
+        private static string AppPath => Path.GetDirectoryName(Environment.GetCommandLineArgs()[0]);
+        private static string TrainDataPath => Path.Combine(AppPath, @"..\..\..\..\datasets\", "taxi-fare-train.csv");
+        private static string TestDataPath => Path.Combine(AppPath, @"..\..\..\..\datasets\", "taxi-fare-test.csv");


We'd like these samples to work in multiple places, I'd think, including cross platform. Do these \ style paths work in Linux?

No, they don't work on Linux.

Instead, we should use Path.Combine("..", "..", "..", "..", "datasets").

TomFinley · 2018-05-22T14:53:11Z

Samples/GettingStarted/BinaryClassification_SentimentAnalysis/Program.cs

+        private static string AppPath => Path.GetDirectoryName(Environment.GetCommandLineArgs()[0]);
+        private static string TrainDataPath => Path.Combine(AppPath, @"..\..\..\..\datasets\", "imdb_labelled.txt");
+        private static string TestDataPath => Path.Combine(AppPath, @"..\..\..\..\datasets\", "yelp_labelled.txt");
+        private static string ModelPath => Path.Combine(AppPath, "Models", "SentimentModel.zip");


"Models", "SentimentModel.zip"); [](start = 65, length = 32)

I've noticed you have I think unintentionally also put the model files (the .zips) in your PR. I think that was unintentional, and is just an artifafct of the run. Either way we have the mechanism to create this artifact here, and we generally try to avoid checking in binary artifacts if it can be helped. Could you remove them?

TomFinley · 2018-05-22T14:58:14Z

Samples/GitHubLabeler/GitHubLabeler/Program.cs

+    {
+        private static async Task Main(string[] args)
+        {
+            if (args.Length != 1)


Is it possible to simplify this sample a bit, so that people don't have to learn some syntax of a simple command line program? As far as I see the sample should just be a straight train of a model, followed by a label, right? Or are we showing how the artifact is preserved across trainings?

Yes, the intention was to demonstrate how it is used in real life where you train it just once and inference many times. Completely agree that for the tutorial samples we want them to be simple train-evaluate-predict. This one is an end-to-end app that you would use for labeling issues. I will probably move it to separate folder for "apps infused with ML" examples.

TomFinley · 2018-05-22T14:59:30Z

Samples/GitHubLabeler/README.md

+## Labeling
+When the model is trained, it can be used for predicting new issue's label. To do so, run the application with `"label"` key:
+```
+C:\GitHubLabeler\GitHubLabeler\bin\Debug\netcoreapp2.0>dotnet GitHubLabeler.dll label


C:\GitHubLabeler\GitHubLabeler\bin\Debug\netcoreapp2.0> [](start = 0, length = 55)

Is it possible to make these samples a bit less Windows specific? That is, perhaps the command to enter should be just on a line somewhere, and you elsewhere say, that it will be placed in such-and-such a directory.

TomFinley · 2018-05-22T15:01:20Z

Samples/GitHubLabeler/README.md

+This is a simple prototype application to demonstrate how to use [ML.NET](https://www.nuget.org/packages/Microsoft.ML/) APIs. The main focus is on creating, training, and using ML (Machine Learning) model that is implemented in Predictor.cs class.
+
+## Overview
+GitHubLabeler is a .NET Core console application that runs from command-line interface (CLI) and allows to:


allows to [](start = 97, length = 9)

Missing pronoun of some sort... allows you to, allows one to, other.

TomFinley · 2018-05-22T15:04:13Z

Samples/GettingStarted/Regression_TaxiFarePrediction/Program.cs

+using Microsoft.ML.Trainers;
+using Microsoft.ML.Transforms;
+using Microsoft.ML;
+using System.Threading.Tasks;


I'm not sure whether we like Microsoft before System or System before Microsoft, but we should probably pick one and make sure it's consistent.

Most developers prefer System go first and have all others alphabetically sorted afterwards.

TomFinley · 2018-05-22T15:06:43Z

Samples/GettingStarted/Regression_TaxiFarePrediction/Program.cs

+
+        private static void Evaluate(PredictionModel<TaxiTrip, TaxiTripFarePrediction> model)
+        {
+            var testData = new TextLoader<TaxiTrip>(TestDataPath, useHeader: true, separator: ",");


var testData = new TextLoader(TestDataPath, useHeader: true, separator: ","); [](start = 12, length = 87)

Something is very wrong here... the loader is part of the model, or ought to be. We shouldn't have to respecify it, any more than we have to respecify the transforms.

I think for evaluation it currently has to be specified, but this should be fixed. See #5 .

I was working on #5 and found out that Loader is actually not part of the trained model in ML.Net. I have opened another issue #216 for this blocker.

TomFinley · 2018-05-22T15:11:04Z

Samples/GitHubLabeler/GitHubLabeler/GitHubLabeler.csproj.DotSettings

@@ -0,0 +1,2 @@
+<wpf:ResourceDictionary xml:space="preserve" xmlns:x="http://schemas.microsoft.com/winfx/2006/xaml" xmlns:s="clr-namespace:System;assembly=mscorlib" xmlns:ss="urn:shemas-jetbrains-com:settings-storage-xaml" xmlns:wpf="http://schemas.microsoft.com/winfx/2006/xaml/presentation">
+	<s:String x:Key="/Default/CodeInspection/CSharpLanguageProject/LanguageLevel/@EntryValue">CSharp72</s:String></wpf:ResourceDictionary>


What's this for?

We shoudn't commit this file as this is from ReSharper ;-)

Will remove it

TomFinley · 2018-05-22T15:19:55Z

Samples/GitHubLabeler/README.md

+
+    b. To work with labels from your GitHub repository, you will need to train the model on your data. To do so, export GitHub issues from your repository into `.tsv` file with the following columns:
+    * ID - issue’s ID
+    * Area - issue’s label (named this way to avoid confusion with the Label concept in ML.NET)


’ [](start = 18, length = 1)

Should we avoid non-ascii characters unless we actually need them? I think I'd prefer a plain-old single quote here.

justinormont · 2018-05-22T15:31:47Z

Samples/GettingStarted/Regression_TaxiFarePrediction/TestTaxiTrips.cs

+            PassengerCount = 1,
+            TripDistance = 10.33f,
+            PaymentType = "CSH",
+            FareAmount = 0 // predict it. actual = 29.5


Do we allow this field to be a null or NaN? It could help the user understand this isn't an input but the output calculated value?

I don't think I can make it nullable type, getting Type mismatch exception.

eerhardt · 2018-05-22T15:45:43Z

datasets/README.md

@@ -0,0 +1,11 @@
+MICROSOFT PROVIDES THE DATASETS ON AN "AS IS" BASIS. MICROSOFT MAKES NO WARRANTIES, EXPRESS OR IMPLIED, GUARANTEES OR CONDITIONS WITH RESPECT TO YOUR USE OF THE DATASETS. TO THE EXTENT PERMITTED UNDER YOUR LOCAL LAW, MICROSOFT DISCLAIMS ALL LIABILITY FOR ANY DAMAGES OR LOSSES, INLCUDING DIRECT, CONSEQUENTIAL, SPECIAL, INDIRECT, INCIDENTAL OR PUNITIVE, RESULTING FROM YOUR USE OF THE DATASETS.


I think it would be good to put all the datasets in a single place - test/data or similar. That way they aren't duplicated unintentionally, and there is one place to see all of the datasets in the repo.

yes, was thinking about it too. I suggest moving datasets from test to the folder datasets in the root and merge them with data that is there for samples. This way it will be obvious for users where to look for it. I can do it if no objections.

There's another competing concern of having too many "things" in the root folder.

A structure I could imagine working would be:

build \ docs \ samples \ datasets \ (or just 'data'?) GettingStarted \ GitHubLabeler \ src \ Project1 \ Project2 \ test \ Test1 \ Test2 \

And all the tests changing to instead of read from test\data, they now point to samples\datasets (or samples\data).

eerhardt · 2018-05-22T15:47:44Z

...Started/BinaryClassification_SentimentAnalysis/BinaryClassification_SentimentAnalysis.csproj

+    <TargetFramework>netcoreapp2.0</TargetFramework>
+  </PropertyGroup>
+
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Debug|AnyCPU'">


This PropertyGroup is incorrect, as the property below won't be set for Release builds.

If you want to set <LangVersion> to latest, it should be done unconditionally. You can move <LangVersion>latest</LangVersion> to the above PropertyGroup.

eerhardt · 2018-05-22T15:48:38Z

Samples/GitHubLabeler/GitHubLabeler/GitHubLabeler.csproj

+    <TargetFramework>netcoreapp2.0</TargetFramework>
+  </PropertyGroup>
+
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Release|AnyCPU'">


Same comment here as above -

Move <LangVersion>latest</LangVersion> into the unconditional property group, and remove these two.

TomFinley · 2018-05-22T15:49:55Z

Hi @OliaG thanks so much for these examples. Let's talk about the size of the example files:

Taxi test: about 25 MB.
Taxi train: about 25 MB.
CoreFX github issue train: about 20 MB.

For machine learning these are pretty small datasets, but they're pretty big as things to commit into a git repo.

Now, we could perhaps do something like utilize Git LFS, which would be an OK remediation except that I see that this PR was done not by working out of a fork then trying to merge back into the main repo, but rather was done by establishing a branch within the main repo itself. So even if we fix it to not have these large files directly, this repo itself is somewhat tainted, since a clone or fork of the repo will contain these large files.

Do we know how to clean this up? I know it's possible, I just don't know how offhand. Maybe @eerhardt ?

Maybe development in forks a good idea, as opposed to branches in the main repo. If we look at the other PRs currently open, they're from branches in people's forks of this repo, not the repo itself. Perhaps we should clarify that somewhere. (Contribution guide, other?)

TomFinley

Hi all, sorry just waiting on changes, especially over large files. I noticed these changes were approved immediately, just want to make sure nothing bad happens.

TomFinley · 2018-05-23T03:14:39Z

datasets/imdb_labelled.txt

@@ -0,0 +1,1000 @@
+A very, very, very slow-moving, aimless movie about a distressed, drifting young man.  	0
+Not sure who was more lost - the flat characters or the audience, nearly half of whom walked out.  	0


Labeled, not labelled, FYI.

It is British VS US spelling. And it is original dataset's name, I did not change it. But I can remove extra "l" :)

When in Rome ...

Oh really? Hmmm. I never knew that. How strange. I guess if that's the original name we should keep it.

* added ceiling to mathutils and have parquetloader default to sequential reading instead of throwing * factor out sequence creation and add check for overflow in MathUtils

…ph variable outputs (#152) * Adding support for training metrics in PipelineSweeperMacro and needed support files. Also includes new output information in PipelineSweeperMacro output graph to make consumption of returned pipelines easier. * Changed where XML comment was placed. * Added more tests (uncommented and fixed) for auto inference. Changed magic number in AutoMlUtils to be mix double value, per review comment. * Added another test, TestPipelineSweeperMacroNoTransforms. * Updated tests to include warning disabling (following Zeeshan S's example) to get build working. * Changes to checks in AutoMlUtils (more correct usage of them). * Fixing issue with ExceptParam using value and not name of parameters. * Fixing errors on use of ExceptParam.

terrajobst · 2018-05-23T16:38:10Z

Samples/GettingStarted/BinaryClassification_SentimentAnalysis/SentimentData.cs

+{
+    public class SentimentData
+    {
+        [Column("0")] public string SentimentText;


I suggest putting the attributes on separate lines as that's more in line with default formatting.

terrajobst · 2018-05-23T16:39:41Z

Samples/GettingStarted/MulticlassClassification_Iris/Program.cs

+    {
+        private static string AppPath => Path.GetDirectoryName(Environment.GetCommandLineArgs()[0]);
+        private static string TrainDataPath => Path.Combine(AppPath, @"..\..\..\..\datasets\", "iris_train.txt");
+        private static string TestDataPath => Path.Combine(AppPath, @"..\..\..\..\datasets\", "iris_test.txt");


Don't use backslashes as this doesn't work cross-platform. You should do:

Path.Combine(AppPath, "..", "..", "..", "..", "datasets", "iris_train.txt");

terrajobst · 2018-05-23T16:40:53Z

Samples/GettingStarted/Regression_TaxiFarePrediction/Program.cs

+using Microsoft.ML.Trainers;
+using Microsoft.ML.Transforms;
+using Microsoft.ML;
+using System.Threading.Tasks;


Most developers prefer System go first and have all others alphabetically sorted afterwards.

terrajobst · 2018-05-23T16:41:44Z

Samples/GitHubLabeler/.gitignore

@@ -0,0 +1,288 @@
+## Ignore Visual Studio temporary files, build results, and
+## files generated by popular Visual Studio add-ons.
+##


Agreed. We should only use the one from the root.

terrajobst · 2018-05-23T16:43:35Z

Samples/GitHubLabeler/GitHubLabeler/App.config

+<?xml version="1.0" encoding="utf-8" ?>
+<configuration>
+  <appSettings>
+    <!--TODO: Please enter your own credentials here. -->


I don't think there is a great solution for this. In general, the recommendation is to avoid using credentials from code and instead use certificates, user-based authentication, or a service like Azure KeyVault. Unfortunately this doesn't work for samples very well...

terrajobst · 2018-05-23T16:44:21Z

Samples/GitHubLabeler/GitHubLabeler/GitHubLabeler.csproj.DotSettings

@@ -0,0 +1,2 @@
+<wpf:ResourceDictionary xml:space="preserve" xmlns:x="http://schemas.microsoft.com/winfx/2006/xaml" xmlns:s="clr-namespace:System;assembly=mscorlib" xmlns:ss="urn:shemas-jetbrains-com:settings-storage-xaml" xmlns:wpf="http://schemas.microsoft.com/winfx/2006/xaml/presentation">
+	<s:String x:Key="/Default/CodeInspection/CSharpLanguageProject/LanguageLevel/@EntryValue">CSharp72</s:String></wpf:ResourceDictionary>


We shoudn't commit this file as this is from ReSharper ;-)

terrajobst · 2018-05-23T16:44:59Z

Samples/GitHubLabeler/README.md

+
+    b. To work with labels from your GitHub repository, you will need to train the model on your data. To do so, export GitHub issues from your repository into `.tsv` file with the following columns:
+    * ID - issue’s ID
+    * Area - issue’s label (named this way to avoid confusion with the Label concept in ML.NET)


* Reduce number of hash bits in stratification column and add a unit test. * Address PR comments.

* replace housing uci dataset to wine quality

Moved datasets from tests to examples Removed model files

- Iris sample - Sentiment Analysis sample - TaxiFare sample -GitHub issues classification sample

Moved datasets from tests to examples Removed model files

…nto samples

KrzysztofCwalina · 2018-05-24T15:10:48Z

This is a very nice PR, but shouldn't we keep all samples/examples in the same place? So far, we have been collecting such samples in tests/Microsoft/ML/Scenarios.

OliaG · 2018-05-24T16:52:22Z

@KrzysztofCwalina regarding keep all samples/example in the same place, here is my understanding.

test/Microsoft/ML/Scenarios are integration tests the primary goal of which is to test our code end-to-end. We can run them to make sure there are no breaking changes introduced by new functionality.

The examples (this PR) are a way of showcasing our functionality to users. They are not tests but independent projects, sometimes applications containing additional logic (like posting requests to GitHub). The examples have to be discoverable for users (like root folder “examples”, it will be hard for users to find them here: test/Microsoft/ML/Scenarios) and F5-ble, users should be able do download just one example and successfully run it. You can’t do it with integration tests.

So for me they belong to different areas and that’s why I would keep tests in tests, and this samples in examples folder. It is a good idea though to mention these tests in docs or readme, so users can find additional use cases.

TomFinley · 2018-05-24T16:57:29Z

docs/code/IdvFileFormat.md

@@ -0,0 +1,191 @@
+# IDV File Format
+
+This document describes ML.NET's Binary dataview file format, version 1.1.1.5


This looks familiar... was this intended to be part of this PR? Was there a bad merge somewhere?

Added GitHubLabeler and GettingStarted

2ff3a88

- Iris sample - Sentiment Analysis sample - TaxiFare sample -GitHub issues classification sample

OliaG requested review from codemzs, eerhardt, terrajobst, asthana86 and shauheen May 22, 2018 05:52

asthana86 approved these changes May 22, 2018

View reviewed changes

asthana86 requested a review from danmoseley May 22, 2018 06:34

justinormont reviewed May 22, 2018

View reviewed changes

shauheen closed this May 22, 2018

shauheen reopened this May 22, 2018

markusweimer reviewed May 22, 2018

View reviewed changes

TomFinley reviewed May 22, 2018

View reviewed changes

justinormont reviewed May 22, 2018

View reviewed changes

eerhardt reviewed May 22, 2018

View reviewed changes

TomFinley suggested changes May 23, 2018

View reviewed changes

TomFinley reviewed May 23, 2018

View reviewed changes

mandyshieh and others added 2 commits May 23, 2018 10:30

Added Block Size Checks for ParquetLoader (#120)

7a5b303

* added ceiling to mathutils and have parquetloader default to sequential reading instead of throwing * factor out sequence creation and add check for overflow in MathUtils

terrajobst suggested changes May 23, 2018

View reviewed changes

yaeldMS and others added 3 commits May 23, 2018 11:30

CV macro with stratification column doesn't work (#213)

73d894b

* Reduce number of hash bits in stratification column and add a unit test. * Address PR comments.

Address review comments

c90ad8a

Fixes build errors caused by spaces in the project path (#196)

76393f4

eerhardt mentioned this pull request May 23, 2018

Taxi fare dataset is almost 50MB #206

Closed

Ivanidzo4ka and others added 8 commits May 23, 2018 16:22

switch housing dataset to wine (#170)

d51321c

* replace housing uci dataset to wine quality

Reorganized datasets

8658ce0

Moved datasets from tests to examples Removed model files

Added GitHubLabeler and GettingStarted

6f0b329

- Iris sample - Sentiment Analysis sample - TaxiFare sample -GitHub issues classification sample

Address review comments

8ac5db8

Reorganized datasets

5ca7cd5

Moved datasets from tests to examples Removed model files

Merge branch 'samples' of https://github.com/dotnet/machinelearning i…

9afcccc

…nto samples

Rebase with master, remove merge conflicts

7cbd7d7

Fixed unit tests

8509d48

Add README for GettingStarted sln

a66ed98

TomFinley reviewed May 24, 2018

View reviewed changes

OliaG closed this May 24, 2018

OliaG mentioned this pull request May 24, 2018

Update samples #238

Closed

justinormont mentioned this pull request Sep 11, 2018

Hot linking to a UCI dataset #889

Closed

ghost locked as resolved and limited conversation to collaborators Mar 30, 2022

		@@ -0,0 +1,2 @@
		<wpf:ResourceDictionary xml:space="preserve" xmlns:x="http://schemas.microsoft.com/winfx/2006/xaml" xmlns:s="clr-namespace:System;assembly=mscorlib" xmlns:ss="urn:shemas-jetbrains-com:settings-storage-xaml" xmlns:wpf="http://schemas.microsoft.com/winfx/2006/xaml/presentation">
		<s:String x:Key="/Default/CodeInspection/CSharpLanguageProject/LanguageLevel/@EntryValue">CSharp72</s:String></wpf:ResourceDictionary>

		@@ -0,0 +1,11 @@
		MICROSOFT PROVIDES THE DATASETS ON AN "AS IS" BASIS. MICROSOFT MAKES NO WARRANTIES, EXPRESS OR IMPLIED, GUARANTEES OR CONDITIONS WITH RESPECT TO YOUR USE OF THE DATASETS. TO THE EXTENT PERMITTED UNDER YOUR LOCAL LAW, MICROSOFT DISCLAIMS ALL LIABILITY FOR ANY DAMAGES OR LOSSES, INLCUDING DIRECT, CONSEQUENTIAL, SPECIAL, INDIRECT, INCIDENTAL OR PUNITIVE, RESULTING FROM YOUR USE OF THE DATASETS.

		@@ -0,0 +1,1000 @@
		A very, very, very slow-moving, aimless movie about a distressed, drifting young man. 0
		Not sure who was more lost - the flat characters or the audience, nearly half of whom walked out. 0

		@@ -0,0 +1,191 @@
		# IDV File Format

		This document describes ML.NET's Binary dataview file format, version 1.1.1.5

Added samples: GitHubLabeler and GettingStarted #198

Added samples: GitHubLabeler and GettingStarted #198

Conversation

OliaG commented May 22, 2018

asthana86 left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

justinormont May 22, 2018 • edited Loading

Choose a reason for hiding this comment

shauheen commented May 22, 2018

markusweimer left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

TomFinley May 22, 2018 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

OliaG May 23, 2018 • edited by terrajobst Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

TomFinley commented May 22, 2018

TomFinley left a comment • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

KrzysztofCwalina commented May 24, 2018

OliaG commented May 24, 2018

Choose a reason for hiding this comment

justinormont May 22, 2018 •

edited

Loading

TomFinley May 22, 2018 •

edited

Loading

OliaG May 23, 2018 •

edited by terrajobst

Loading

TomFinley left a comment •

edited

Loading