Main ReadMe, November 21, 2016

README.md

@@ -3,7 +3,7 @@
 Highlights of this Release:
 * New ASGD/Hogwild! training using Microsoft’s Parameter Server ([Project Multiverso](https://github.com/Microsoft/multiverso))
 * Distributed Scenarios now supported in CNTK Python API
-* New Memory compression -- ability to trade off memory usage with compute. 
+* Introducing of [Memory compression](https://github.com/Microsoft/CNTK/wiki/Top-level-configurations#hypercompressmemory) optimizing memory usage, especially for GPU computation 
 * CNTK Docker image with 1bit-SGD support
 * Stability Improvements and bug fixes
 
@@ -55,18 +55,6 @@ If you ARE using Model Evaluation Library we **strongly recommend** installing v
 
 See [Release Notes](https://github.com/Microsoft/CNTk/wiki/CNTK_1_7_2_Release_Notes) for details.
 
-*2016-09-28.* V 1.7.1 Binary release  
-Highlights of this Release:
-* Two Breaking Changes related to Layers library default initialization and ```fsAdagrad``` gradient-normalization scheme
-* Improvements in BrainScript
-* Enabling of Deterministic Algorithm enforcement
-* Improvements in Model Evaluation including the support of Evaluation for Azure Applications
-* Different Performance improvements
-* Multiple bug fixes
-
-See more in the [Release Notes](https://github.com/Microsoft/CNTK/wiki/CNTK_1_7_1_Release_Notes) (including the full list of bugs fixed)  
-Get the Release from the [CNTK Releases page](https://github.com/Microsoft/CNTK/releases)
-
 See [all news](https://github.com/Microsoft/CNTK/wiki/News).
 
 # What is CNTK