Add olmOCR-bench old_scans result (community)
Browse filesCommunity-provided eval: PaddleOCR-VL-1.6 on olmOCR-bench old_scans = 38.6 (0-100, old_scans.jsonl sub-score: present/absent/order). Default v1.6 pipeline, greedy, no tuning. Harness validated by reproducing the published original-PaddleOCR-VL figure (37.8 -> 38.2, within CI). Method + scripts: https://github.com/davanstrien/ocr-bench/tree/d07a41232/experiments/olmocr-bench-oldscans
.eval_results/olmocrbench.yaml
ADDED
|
@@ -0,0 +1,10 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
- dataset:
|
| 2 |
+
id: allenai/olmOCR-bench
|
| 3 |
+
task_id: old_scans
|
| 4 |
+
value: 38.6
|
| 5 |
+
date: "2026-06-27"
|
| 6 |
+
source:
|
| 7 |
+
url: https://github.com/davanstrien/ocr-bench/tree/d07a41232/experiments/olmocr-bench-oldscans
|
| 8 |
+
name: ocr-bench — old_scans eval
|
| 9 |
+
user: davanstrien
|
| 10 |
+
notes: "PaddleOCR-VL-1.6 pipeline, greedy, no tuning; old_scans.jsonl sub-score"
|