|
| 1 | +# [ML.NET](http://dot.net/ml) 4.0-preview1 |
| 2 | + |
| 3 | +## **New Features** |
| 4 | +- **Add sweepable estimator to NER** ([6965](https://github.com/dotnet/machinelearning/pull/6965)) |
| 5 | +- **Introducing Tiktoken Tokenizer** ([6981](https://github.com/dotnet/machinelearning/pull/6981)) |
| 6 | +- **Add text normalizer transformer to AutoML** ([6998](https://github.com/dotnet/machinelearning/pull/6998)) |
| 7 | + |
| 8 | +## **Enhancements** |
| 9 | +- **Add support for Apache.Arrow.Types.TimestampType to DataFrame** ([6871](https://github.com/dotnet/machinelearning/pull/6871)) - Thanks @asmirnov82! |
| 10 | +- **Add new type to key-value converter** ([6973](https://github.com/dotnet/machinelearning/pull/6973)) |
| 11 | +- **Update OnnxRuntime to 1.16.3** ([6975](https://github.com/dotnet/machinelearning/pull/6975)) |
| 12 | +- **Tokenizer's Interfaces Cleanup** ([7001](https://github.com/dotnet/machinelearning/pull/7001)) |
| 13 | +- **Match SweepableEstimatorFactory name with Ml.net name.** ([7007](https://github.com/dotnet/machinelearning/pull/7007)) |
| 14 | +- **First round of perf improvements for tiktoken** ([7012](https://github.com/dotnet/machinelearning/pull/7012)) |
| 15 | +- **Tweak CreateByModelNameAsync** ([7015](https://github.com/dotnet/machinelearning/pull/7015)) |
| 16 | +- **Avoid LruCache in Tiktoken when cacheSize specified is 0** ([7016](https://github.com/dotnet/machinelearning/pull/7016)) |
| 17 | +- **Tweak Tiktoken's BytePairEncode for improved perf** ([7017](https://github.com/dotnet/machinelearning/pull/7017)) |
| 18 | +- **Optimize regexes used in tiktoken** ([7020](https://github.com/dotnet/machinelearning/pull/7020)) |
| 19 | +- **Address the feedback on the tokenizer's library** ([7024](https://github.com/dotnet/machinelearning/pull/7024)) |
| 20 | +- **Add Span support in tokenizer's Model abstraction** ([7035](https://github.com/dotnet/machinelearning/pull/7035)) |
| 21 | +- **Adding needed Tokenizer's APIs** ([7047](https://github.com/dotnet/machinelearning/pull/7047)) |
| 22 | + |
| 23 | +## **Bug Fixes** |
| 24 | +- **Fix formatting that fails in VS** ([7023](https://github.com/dotnet/machinelearning/pull/7023)) |
| 25 | +- **Issue #6606 - Add sample variance and standard deviation to NormalizeMeanVariance** ([6885](https://github.com/dotnet/machinelearning/pull/6885)) - Thanks @tearlant! |
| 26 | + |
| 27 | +## **Build / Test updates** |
| 28 | +- **Migrate to the 'locker' GitHub action for locking closed/stale issues/PRs** ([6896](https://github.com/dotnet/machinelearning/pull/6896)) |
| 29 | +- **Reorganize dataframe files** ([6872](https://github.com/dotnet/machinelearning/pull/6872)) - Thanks @asmirnov82! |
| 30 | +- **Updated ml.net versioning** ([6907](https://github.com/dotnet/machinelearning/pull/6907)) |
| 31 | +- **Don't include the SDK in our helix payload** ([6918](https://github.com/dotnet/machinelearning/pull/6918)) |
| 32 | +- **Make double assertions compare with tolerance instead of precision** ([6923](https://github.com/dotnet/machinelearning/pull/6923)) |
| 33 | +- **Fix assert by only accessing idx** ([6924](https://github.com/dotnet/machinelearning/pull/6924)) |
| 34 | +- **Only use semi-colons for NoWarn - fixes build break** ([6935](https://github.com/dotnet/machinelearning/pull/6935)) |
| 35 | +- **Packaging cleanup** ([6939](https://github.com/dotnet/machinelearning/pull/6939)) |
| 36 | +- **Add Backport github workflow** ([6944](https://github.com/dotnet/machinelearning/pull/6944)) |
| 37 | +- **[main] Update dependencies from dotnet/arcade** ([6957](https://github.com/dotnet/machinelearning/pull/6957)) |
| 38 | +- **Update .NET Runtimes to latest version** ([6964](https://github.com/dotnet/machinelearning/pull/6964)) |
| 39 | +- **Testing light gbm bad allocation** ([6968](https://github.com/dotnet/machinelearning/pull/6968)) |
| 40 | +- **[main] Update dependencies from dotnet/arcade** ([6969](https://github.com/dotnet/machinelearning/pull/6969)) |
| 41 | +- **[main] Update dependencies from dotnet/arcade** ([6976](https://github.com/dotnet/machinelearning/pull/6976)) |
| 42 | +- **FabricBot: Onboarding to GitOps.ResourceManagement because of FabricBot decommissioning** ([6983](https://github.com/dotnet/machinelearning/pull/6983)) |
| 43 | +- **[main] Update dependencies from dotnet/arcade** ([6985](https://github.com/dotnet/machinelearning/pull/6985)) |
| 44 | +- **[main] Update dependencies from dotnet/arcade** ([6995](https://github.com/dotnet/machinelearning/pull/6995)) |
| 45 | +- **Temp fix for the race condition during the tests.** ([7021](https://github.com/dotnet/machinelearning/pull/7021)) |
| 46 | +- **Make MlImage tests not block file for reading** ([7029](https://github.com/dotnet/machinelearning/pull/7029)) |
| 47 | +- **Remove SourceLink SDK references** ([7037](https://github.com/dotnet/machinelearning/pull/7037)) |
| 48 | +- **Change official build to use 1ES templates** ([7048](https://github.com/dotnet/machinelearning/pull/7048)) |
| 49 | +- **Auto-generated baselines by 1ES Pipeline Templates** ([7051](https://github.com/dotnet/machinelearning/pull/7051)) |
| 50 | +- **Update package versions in use by ML.NET tests** ([7055](https://github.com/dotnet/machinelearning/pull/7055)) |
| 51 | +- **testing arm python brew overwite** ([7058](https://github.com/dotnet/machinelearning/pull/7058)) |
| 52 | + |
| 53 | +## **Documentation Updates** |
| 54 | +- **Update developer-guide.md** ([6870](https://github.com/dotnet/machinelearning/pull/6870)) - Thanks @computerscienceiscool! |
| 55 | +- **Update release-3.0.0.md** ([6895](https://github.com/dotnet/machinelearning/pull/6895)) - Thanks @taeerhebend! |
0 commit comments