docs: add initial version

README.md

@@ -0,0 +1,13 @@
+---
+license: apache-2.0
+language:
+- de
+---
+
+# nanochat German: Pretraining Data
+
+This repository uses a subset of the [LLäMmlein](https://huggingface.co/datasets/LSX-UniWue/LLaMmlein-Dataset) pretraining dataset, which itself is a strict subset of the German portion of the [RedPajama V2](https://huggingface.co/datasets/togethercomputer/RedPajama-Data-V2) dataset.
+
+To construct the dataset, we download the first 20 JSONL files from LLäMmlein and merge them into a single collection.
+
+Following the original nanochat dataset construction process, we then split the data into shards containing approximately 250 million characters each.