Update README.md

A collection 1M batches of 64 Llama 3 8B inferencing GEMMs, randomly selected. The values for N in the GEMMs were selected by selecting from the distribution of token lengths in the shareGPT dataset.

README.md

@@ -16,4 +16,5 @@ configs:
   data_files:
   - split: train
     path: data/train-*
----
+license: mit
+---