Update README.md

README.md

@@ -5,7 +5,7 @@ license: other
 See LICENSE file for license.
 This is a collection of merged, then converted to 4bit LLaMA models trained on the storytelling dataset I used for the storytelling LoRAs.
 
-Unlike the LoRAs, some formatting oddness seems to have broken through.
-Triple newlines tend to start new chapters which can break flow.
+UPDATE: 04/04
+Cleaned data and retrained to 32 groupsize and safetensors. Formatting oddities seem to have been wiped out.
 
-A 30B model was converted too, but formatting is very broken in that case, for some unknown reason.
+Format: Nothing, paragraphs separated by ""***"".