Updated documentation

README.md

@@ -1,4 +1,34 @@
 # Tsetlin Machine Unified - One Codebase to Rule Them All
 ![License](https://img.shields.io/github/license/microsoft/interpret.svg?style=flat-square) ![Python Version](https://img.shields.io/pypi/pyversions/interpret.svg?style=flat-square)![Maintenance](https://img.shields.io/maintenance/yes/2023?style=flat-square)
 
-Implements the Tsetlin Machine (https://arxiv.org/abs/1804.01508), Coalesced Tsetlin Machine (https://arxiv.org/abs/2108.07594), Convolutional Tsetlin Machine (https://arxiv.org/abs/1905.09688), Regression Tsetlin Machine (https://royalsocietypublishing.org/doi/full/10.1098/rsta.2019.0165), and Weighted Tsetlin Machine (https://ieeexplore.ieee.org/document/9316190), with support for continuous features (https://arxiv.org/abs/1905.04199), drop clause (https://arxiv.org/abs/2105.14506), Type III Feedback (to be published), focused negative sampling (https://ieeexplore.ieee.org/document/9923859), multi-task classifier (to be published), autoencoder (https://arxiv.org/abs/2301.00709), literal budget (to be published), incremental clause evaluation (to be published), and one-vs-one multi-class classifier (to be published). TMU is written in Python with wrappers for C and CUDA-based clause evaluation and updating.
+The TMU repository is a collection of Tsetlin Machine implementations, namely:
+* Tsetlin Machine (https://arxiv.org/abs/1804.01508)
+* Coalesced Tsetlin Machine (https://arxiv.org/abs/2108.07594)
+* Convolutional Tsetlin Machine (https://arxiv.org/abs/1905.09688)
+* Regression Tsetlin Machine (https://royalsocietypublishing.org/doi/full/10.1098/rsta.2019.0165)
+* Weighted Tsetlin Machine (https://ieeexplore.ieee.org/document/9316190)
+
+Further, we implement many TM features, including:
+* Support for continuous features (https://arxiv.org/abs/1905.04199)
+* Drop clause (https://arxiv.org/abs/2105.14506)
+* Type III Feedback (to be published)
+* Focused negative sampling (https://ieeexplore.ieee.org/document/9923859)
+* Multi-task classifier (to be published)
+* Autoencoder (https://arxiv.org/abs/2301.00709)
+* Literal budget (to be published)
+* Incremental clause evaluation (to be published)
+* One-vs-one multi-class classifier (to be published).
+
+TMU is written in Python with wrappers for C and CUDA-based clause evaluation and updating.
+
+# Installation
+
+## Installing on Windows
+To install on windows, you will need the MSVC build tools, (https://visualstudio.microsoft.com/visual-cpp-build-tools/
+)[found here].  When prompted, select the `Workloads → Desktop development with C++` package, 
+which is roughly 6-7GB of size, install it and you should be able to compile the cffi modules.
+
+## Installing TMU
+```bash
+pip install git+https://github.com/cair/tmu.git
+```