Upload README.md with huggingface_hub

README.md

@@ -1,16 +1,16 @@
 ---
 language:
-- en
-- th
+  - en
+  - th
 tags:
-- transliteration
-- thai
-- english
-- byt5
-- fp16
+  - transliteration
+  - thai
+  - english
+  - byt5
+  - fp16
 license: mit
 datasets:
-- custom
+  - custom
 library_name: transformers
 pipeline_tag: text2text-generation
 ---
@@ -63,9 +63,9 @@ The model was trained on a combined dataset of English-Thai transliteration pair
   - Gradient clipping: Yes (max_grad_norm=1.0)
 
 ### Evaluation Results
-- **Accuracy**: `0.7562`
-- **Character Error Rate**: `0.0568`
-- **Mean Levenshtein Distance**: `0.4473`
+- **Accuracy**: `0.7831`
+- **Character Error Rate**: `0.0591`
+- **Mean Levenshtein Distance**: `0.4654`
 
 ## How to Use
 
@@ -148,9 +148,8 @@ The model's performance may vary based on:
 ## Common Errors
 
 Some common error patterns observed:
-- greenockite → กรีนอกไทต์ (should be: กรีนอกไคต์)
-- group → กรุ๊ป (should be: กรูป)
-- glauconite → เกลาโคไนต์ (should be: กลอโคไนต์)
+- group → กรูป (should be: กรุ๊ป)
+- golf → โกล์ฟ (should be: กอล์ฟ)
 
 ## License