Update README.md

README.md

@@ -7,4 +7,4 @@ quantized_by: kromeurus
 
 For the details of this model, please refer to the orginal model card [here](https://huggingface.co/kromeurus/L3.1-Tivir-v0.1-10B).
 
-Additionally, there is another Q_8 quant in this repo named 'fQ_8'. It's not a new type of quant, rather a Q_8 quant that has the full float32 tokenizer. It's slightly larger then a normal Q_8 and meant to be an alternative maybe better quality quant. 
+Additionally, there is another Q_8 quant in this repo named 'fQ_8'. It's not a new type of quant, rather a Q_8 quant that has the full float32 tokenizer. It's slightly larger then a normal Q_8 and meant to be an alternative quant. If it's giving you shitty outputs, use the normal Q_8.