Release notes for 1.7.0 preview (#5982)

Author: michaelgsharp

docs/release-notes/1.7.0/release-1.7.0.md

@@ -0,0 +1,41 @@
+# [ML.NET](http://dot.net/ml) 1.7.0
+
+Moving forward, we are going to be aligning more with the overall .NET release schedule. As such, this is a smaller release since we had a larger one just about 3 months ago but it aligns us with the release of .NET 6.
+
+## **New Features**
+### ML.NET
+- **Switched to getting version from assembly custom attributes**- ([#4512](https://github.com/dotnet/machinelearning/pull/4512)) Remove reliance on getting product version for model.zip/version.txt from FileVersionInfo and replace with using assembly custom attributes. This will help in supporting single file applications. (**Thanks @r0ss88**)
+- **Can now optionally not dispose of the underlying model when you dispose a prediction engine**. ([#5964](https://github.com/dotnet/machinelearning/pull/5964)) A new prediction engine options class has been added that lets you determine if the underlying model should be disposed of or not when the prediction engine itself is disposed of.
+- **Can now set the number of threads that onnx runtime uses** ([#5962](https://github.com/dotnet/machinelearning/pull/5962)) This lets you specify the number of parallel threads ONNX runtime will use to execute the graph and run the model. (**Thanks @yaeldekel**)
+- **The PFI API has been completely reworked and is now much more user friendly** ([#5934](https://github.com/dotnet/machinelearning/pull/5934)) You can now get the output from PFI as a dictionary mapping the column name (or the slot name) to its PFI result.
+### DataFrame
+- **Can now merge using multiple columns in a JOIN condition** ([#5838](https://github.com/dotnet/machinelearning/pull/5838)) (**Thanks @asmirnov82**)
+
+## **Enhancements**
+### ML.NET
+- Run formatting on all src projects ([#5937](https://github.com/dotnet/machinelearning/pull/5937)) (**Thanks @jwood803**)
+- Added BufferedStream for reading from DeflateStream - reduces loading time for .NET core ([#5924](https://github.com/dotnet/machinelearning/pull/5924)) (**Thanks @martintomasek**)
+- Update editor config to match Roslyn and format samples ([#5893](https://github.com/dotnet/machinelearning/pull/5893)) (**Thanks @jwood803**)
+- Few more minor editor config changes ([#5933](https://github.com/dotnet/machinelearning/pull/5933))
+### DataFrame
+- Use Equals and = operator for DataViewType comparison ([#5942](https://github.com/dotnet/machinelearning/pull/5942)) (**Thanks @thoron**)
+
+
+
+## **Bug Fixes**
+- Initialize _bestMetricValue when using the Loss metric ([#5939](https://github.com/dotnet/machinelearning/pull/5939)) (**Thanks @MiroslavKabat**)
+
+
+## **Build / Test updates**
+- Changed the queues used for building/testing from Ubuntu 16.04 to 18.04 ([#5970](https://github.com/dotnet/machinelearning/pull/5970))
+- Add in support for building with VS 2022. ([#5956](https://github.com/dotnet/machinelearning/pull/5956))
+- Codecov yml token was added ([#5950](https://github.com/dotnet/machinelearning/pull/5950))
+- Move from XliffTasks to Microsoft.DotNet.XliffTasks ([#5887](https://github.com/dotnet/machinelearning/pull/5887))
+
+
+## **Documentation Updates**
+- Fixed up Readme, updated the roadmap, and new doc detailing some platform limitations. ([#5892](https://github.com/dotnet/machinelearning/pull/5892))
+
+
+## **Breaking Changes**
+- None