Add evaluation section

README.md

@@ -240,6 +240,13 @@ This corpus is intended for:
 - Language modeling and NLP tasks for Armenian  
 - OCR post-processing and text normalization research
 
+## Evaluation
+
+Predictions against this dataset are scored with the evaluation kit in
+[`evaluation_kit/`](https://huggingface.co/datasets/RLALT/ACoPPer/tree/main/evaluation_kit)
+(in particular [`evaluate_from_hf.py`](https://huggingface.co/datasets/RLALT/ACoPPer/blob/main/evaluation_kit/evaluate_from_hf.py),
+run with `--dataset RLALT/ACoPPer`).
+
 ## OCR Evaluation Results
 
 This section tracks the Character Error Rate (CER) across various baseline architectures and configuration filters on the `ACoPPer` test split. Note that for CER, **lower values indicate better performance**.