Update README.md

README.md

@@ -1,77 +1,63 @@
 ---
 language:
-  - bo
+- bo
 license: apache-2.0
 tags:
-  - image-classification
-  - tibetan
-  - uchen
-  - ume
-  - script-classification
-  - dinov3
-  - fine-tuned
+- image-classification
+- tibetan
+- uchen
+- ume
+- script-classification
+- dinov3
+- fine-tuned
 library_name: transformers
 pipeline_tag: image-classification
 base_model: facebook/dinov3-vits16-pretrain-lvd1689m
 datasets:
-  - openpecha/uchen-ume-classification-benchmark
+- openpecha/uchen_ume_classification_dataset
 metrics:
-  - f1
-  - accuracy
+- f1
+- accuracy
 model-index:
-  - name: Uchen-Ume Classifier (DINOv3 ViT-S) — center crop
-    results:
-      - task:
-          type: image-classification
-          name: Tibetan Script Classification (center-crop whole page)
-        dataset:
-          name: openpecha/uchen-ume-classification-benchmark
-          type: openpecha/uchen-ume-classification-benchmark
-          split: test
-        metrics:
-          - name: Macro F1 (center crop)
-            type: f1
-            value: 0.983
-          - name: Accuracy (center crop)
-            type: accuracy
-            value: 0.993
-  - name: Uchen-Ume Classifier (DINOv3 ViT-S) — full page
-    results:
-      - task:
-          type: image-classification
-          name: Tibetan Script Classification (full page)
-        dataset:
-          name: openpecha/uchen-ume-classification-benchmark
-          type: openpecha/uchen-ume-classification-benchmark
-          split: test
-        metrics:
-          - name: Macro F1 (full page)
-            type: f1
-            value: 0.708
-          - name: Accuracy (full page)
-            type: accuracy
-            value: 0.807
-      - task:
-          type: image-classification
-          name: Held-out benchmark (full page)
-        dataset:
-          name: openpecha/uchen-ume-classification-benchmark
-          type: openpecha/uchen-ume-classification-benchmark
-          split: benchmark
-        metrics:
-          - name: Macro F1 (full page)
-            type: f1
-            value: 0.848
-          - name: Accuracy (full page)
-            type: accuracy
-            value: 0.850
+- name: Uchen-Ume Classifier (DINOv3 ViT-S) — center crop
+  results:
+  - task:
+      type: image-classification
+      name: Tibetan Script Classification (center-crop whole page)
+    dataset:
+      name: openpecha/uchen-ume-classification-benchmark
+      type: openpecha/uchen-ume-classification-benchmark
+      split: test
+    metrics:
+    - name: Macro F1 (center crop)
+      type: f1
+      value: 0.983
+    - name: Accuracy (center crop)
+      type: accuracy
+      value: 0.993
+- name: Uchen-Ume Classifier (DINOv3 ViT-S) — full page
+  results:
+  - task:
+      type: image-classification
+      name: Tibetan Script Classification (full page)
+    dataset:
+      name: openpecha/uchen-ume-classification-benchmark
+      type: openpecha/uchen-ume-classification-benchmark
+      split: test
+    metrics:
+    - name: Macro F1 (full page)
+      type: f1
+      value: 0.708
+    - name: Accuracy (full page)
+      type: accuracy
+      value: 0.807
 ---
 
 # Uchen vs Umê Classifier (DINOv3 ViT-S)
 
 Binary Tibetan script classifier: **Uchen** (དབུ་ཅན།, headed/printed script) vs **Umê** (དབུ་མེད།, headless/cursive script). Fine-tuned from [DINOv3 ViT-S](https://huggingface.co/facebook/dinov3-vits16-pretrain-lvd1689m) on ~10,000 manuscript scans from the [Buddhist Digital Resource Center](https://www.bdrc.io) (BDRC).
 
-**Dataset:** [openpecha/uchen-ume-classification-benchmark](https://huggingface.co/datasets/openpecha/uchen-ume-classification-benchmark)
+**Dataset:** [openpecha/uchen-ume-classification-dataset](https://huggingface.co/datasets/openpecha/uchen-ume-classification-dataset)
 
 ## Which checkpoint to use
 
@@ -86,7 +72,7 @@ Pick the variant that matches **how you preprocess at inference**:
 
 ## Best results
 
-Hub split: 9,110 train / 1,000 val / 851 test (work-stratified). Benchmark holdout: 60 pages.
+Hub split: 9,110 train / 1,000 val / 851 test (work-stratified).
 
 | Variant | Train | Val | Test @ eval | Test acc | Test macro-F1 | Val macro-F1 (best) |
 |---------|-------|-----|-------------|:--------:|:-------------:|:-------------------:|
@@ -94,15 +80,6 @@ Hub split: 9,110 train / 1,000 val / 851 test (work-stratified). Benchmark holdo
 | **`without_preprocess/`** | none | none | none (full page) | **80.7%** | **0.708** | 0.771 |
 | `with_preprocess/` (legacy) | center crop | center crop | full page | 56.1% | 0.506 | 0.994 |
 
-### Full-page benchmark (60 pages, `preprocess none`)
-
-| Variant | Benchmark acc | Benchmark macro-F1 |
-|---------|:-------------:|:------------------:|
-| `without_preprocess/` | **85.0%** | **0.848** |
-| `with_preprocess/` | 68.3% | 0.648 |
-
-Run benchmark eval for `center_crop_all/` with `--preprocess center_crop_whole_page` to match training.
-
 ### Test confusion matrices (851 pages)
 
 | Variant | uchen→uchen | uchen→ume | ume→uchen | ume→ume |
@@ -165,8 +142,6 @@ path = hf_hub_download(
 
 ```bash
 python inference_uchen_ume.py \
-  --benchmark-json benchmark/benchmark_holdout.json \
-  --fetch-urls \
   --weights without_preprocess/final_model.pt \
   --preprocess none
 ```
@@ -180,15 +155,12 @@ center_crop_all/             ← center_crop_whole_page at inference (~99% test)
   results.json               ← includes confusion_matrix
   confusion_matrix.json
   confusion_matrix.png
-without_preprocess/          ← full pages (~81% test, ~85% benchmark)
+without_preprocess/          ← full pages (~81% test)
   final_model.pt
   model_card.json
   results.json
   confusion_matrix.json
   confusion_matrix.png
-  benchmark_eval_results.json   ← benchmark CM in JSON
-with_preprocess/             ← legacy mismatch — do not use
-  ...
 ```
 
 ## Limitations
@@ -201,15 +173,15 @@ with_preprocess/             ← legacy mismatch — do not use
 
 ```bibtex
 @misc{karma2026uchenume,
-    title   = {Uchen-Ume Classifier: Binary Tibetan Script Classification with DINOv3},
-    author  = {Karma Tashi and Elie Roux},
-    year    = {2026},
-    url     = {https://huggingface.co/openpecha/uchen-ume-classifier},
-    note    = {Fine-tuned on openpecha/uchen-ume-classification-benchmark.
-               Funded by Khyentse Foundation. Images from BDRC.}
+    title        = {Uchen-Ume Classifier: Binary Tibetan Script Classification with DINOv3},
+    author       = {Karma Tashi and Elie Roux},
+    year         = {2026},
+    publisher    = {HuggingFace},
+    url          = {https://huggingface.co/openpecha/uchen-ume-classifier},
+    note         = {Funded by Khyentse Foundation. Images sourced from the Buddhist Digital Resource Center (BDRC).}
 }
 ```
 
 ## Acknowledgements
 
-Developed by **Dharmaduta** for the **[Buddhist Digital Resource Center](https://www.bdrc.io)** (BDRC) Etext Corpus project, with funding from the **Khyentse Foundation**. Annotation guidelines by **Pentsok Rtsang**.
+Developed by **Dharmaduta** for the **[Buddhist Digital Resource Center](https://www.bdrc.io)** (BDRC) Etext Corpus project, with funding from the **Khyentse Foundation**. Annotation guidelines by **Pentsok Rtsang**.