Update README.md

README.md

@@ -79,6 +79,13 @@ The subsets can be for this training step are derived from 2 different sources:
 - Enhancement: Synthetic data generation by Avemio AG
 - Quality: Automatic validation and curation of examples by Open Source LLM's
 
+### Methodology: Reasoning-DE & Reasoning-EN
+- Providing Persona Descriptions and rewriting in a similar style with a different focus area and name in german/english language
+- Generating Simple Logical Problems out of Persona-specific Views & Language.
+- Generating Approaches, Thinking-Steps & Solutions separately verified by Llama-3.1-70B-Instruct
+- Structure addition for specific task types
+- Quality assurance and validation
+
 ## Task Types and Examples
 
 ### 1. Classification JSON