Added note on repetition penalty setting

README.md

@@ -9,6 +9,8 @@ datasets:
 # Fine-tune of Y-34B with Spicyboros-3.1
 One epoch of fine tuning with @jondurbin's SpicyBoros-3.1 dataset.  4.65bpw should fit on a single 3090/4090, 5.0bpw, 6.0bpw, and 8.0bpw will require more than one GPU 24 GB VRAM GPU.
 
+**Please note:** you may have to turn down repetition penalty to 1.0. The model seems to get into "thesaurus" mode sometimes without this change.
+
 # Original Yi-34B Model Card Below
 <div align="center">
 
@@ -54,4 +56,4 @@ Although we use data compliance checking algorithms during the training process
 ## License
 
 The Yi series model must be adhere to the [Model License Agreement](https://huggingface.co/01-ai/Yi-34B/blob/main/LICENSE).
-For any questions related to licensing and copyright, please contact us ([[email protected]](mailto:[email protected])).
+For any questions related to licensing and copyright, please contact us ([[email protected]](mailto:[email protected])).