Update README.md

README.md

@@ -25,18 +25,25 @@ This page confirms the effectiveness of multilingual imatrix.
 ![wiki.test.raw_perplexity_score.png](wiki.test.raw_perplexity_score.png)
 
 
-![wiki.test.raw_perplexity_score.png](wiki.test.raw_perplexity_score.png)
+![ja-wiki.test.raw_perplexity_score.png](ja-wiki.test.raw_perplexity_score.png)
 
 
 
 #### Terminology
 
-Bartowski model  
-Bartowski is a person who has quantized many models and contributed to the community. He creates imatrix from English-only data[calibration_datav3.txt](https://gist.github.com/bartowski1182/eb213dccb3571f863da82e99418f81e8).  
+wiki.test.raw score  
+Perplexity Score measured using wiki.test.raw published by Salesforce and the llama-perplexity command with -c 512 setting.  
+
+ja-wiki.test.raw perplexity score  
+Perplexity score measured in the same way using Japanese data ([ja-wiki.test.raw]()) with the same file size as wiki.test.raw, extracted from a Japanese wiki.  
 
-Imatrix-jpn-test model
 
-Comparing the performance of models created with the multilingual version of Imatrix
+
+Bartowski model  
+[Bartowski](https://huggingface.co/bartowski) is a person who has quantized many models and contributed to the community. He creates imatrix from English-only data[calibration_datav3.txt](https://gist.github.com/bartowski1182/eb213dccb3571f863da82e99418f81e8).  
+
+Imatrix-jpn-test model  
+This model. Its created with the multilingual version of text[calibration_datav3_plus_jpn_v1.txt](https://huggingface.co/dahara1/imatrix-jpn-test/blob/main/calibration_datav3_plus_jpn_v1.txt).