Dr.DocBench Challenge

Overview

The Dr.DocBench Challenge evaluates expert-level document parsing on pages where strong systems still struggle. It is based on Dr.DocBench, a difficulty-aware benchmark developed by 2077AI and research collaborators, and is part of the DocInsights 2026 workshop at EMNLP 2026.

Challenge Task

Expert-Level Document Parsing

Participants convert difficult document images into structured Markdown while preserving:

Original-language text Structured HTML tables LaTeX formulas Reading order Document structure Chemistry notation Musical notation (exploratory content)

Musical notation appears in the challenge data as exploratory content. It does not receive a separate EvalAI metric.

Timeline

August 10, 2026

Competition opens

October 10, 2026 · 12:59 PM UTC

Submission deadline

October 11–23, 2026

Final evaluation period

DocInsights 2026

Winners announced at the workshop

Awards

US$5,000 Combined Prize Pool

Shared across the Dr. DocBench and DocSEM challenges. Award details and allocation will be announced soon.

Leaderboard

Current standings · provisional

This is not the final leaderboard. Scores and rankings may change as new submissions are evaluated, and the official final results will be published after the competition ends.

1

Current leader

ohmyyuan

82.63Overall

2

Second place

pranshu29

81.59Overall

3

Third place

ppppppp

80.50Overall

Expert-Level Document Parsing

Last updated Sunday, Sep 27, 2026

Rank	Team
1	ohmyyuan	0.1289	63.4200	35.0600	82.3300	82.6300
2	pranshu29	0.1681	68.0000	55.0000	82.8500	81.5900
3	ppppppp	0.1692	71.8000	33.4700	82.3700	80.5000
4	21sxs	0.1384	73.4200	0.5000	82.4300	80.2700
5	ccpcpcp	0.1400	73.1300	0.4400	82.2200	80.1400
6	cccppp	0.1445	72.8600	0.4400	81.7400	79.7900
7	Navidc-OCR	0.1464	72.8600	0.4400	81.5200	79.5900
8	qwerdas	0.1464	72.8600	0.4400	81.5200	79.5900
9	kepa	0.1465	72.8700	0.0000	81.5200	79.5700
10	TOP CHAi	0.1954	69.8800	24.0700	83.2100	78.8300
11	Govind	0.1910	66.6800	0.2600	83.5900	77.7600
12	AIKHEMIST	0.1673	66.4200	0.0300	78.6300	76.7800
13	hero989 (s21 = s20 + stage-direction fo)	0.2143	63.8400	5.3300	81.4700	75.9900
14	PSK	0.1999	62.2600	0.0000	80.6200	75.9100
15	PageWeaver	0.2380	54.1800	0.0400	81.2500	74.5500
16	OC	0.1789	71.3900	0.0000	71.5000	72.9700
17	tranquil	0.1823	71.3900	0.0000	71.4200	72.8200
18	xyzxyzxyz	0.2170	70.7700	31.7900	69.8300	72.3300
19	docmarvel	0.2036	70.1300	0.0000	70.5200	71.2800
20	Team-Infrrd	0.2089	67.9000	0.1300	69.3400	70.5300
21	agentereal	0.2766	58.1800	0.2200	75.0600	70.3800
22	Stockmark	0.2322	52.5400	0.0000	71.8100	70.1600
23	ChrisTeam	0.3126	47.6300	62.6500	72.1900	70.1300
24	WPS-MonkeyOCR	0.3049	63.7600	0.1600	77.9400	69.8000
25	ddun	0.2497	49.8800	0.8400	71.2000	69.0900
26	krs	0.2429	61.9500	0.1300	69.7400	69.0600
27	blue0620	0.2205	66.8900	0.3700	66.6800	68.8400
28	F8FQ	0.3230	66.8700	0.0400	75.3500	68.1200
29	DocTA	0.2372	73.8000	2.1700	61.4500	66.4700
30	CPR	0.2735	60.9100	0.8400	64.0500	65.4500
31	alternative	0.2543	53.4800	0.0000	61.1000	64.8900
32	clas	0.2893	54.3600	0.0000	65.4700	64.6000
33	Docsly	0.2931	48.2300	0.0000	62.4400	63.3700
34	we'll never kniw	0.3345	67.4900	0.0000	65.1900	62.9500
35	PrizeHunter (exp_9025 S=80.0101 [human_choi)	0.2903	16.3500	0.3200	61.5300	62.3700
36	Ron7777	0.3235	68.2900	0.0000	61.5300	62.1500
37	zoe	0.3409	67.4900	0.0800	64.0200	62.0800
38	xuhai-docbench-2717	0.3367	66.4700	0.3300	63.6200	62.0600
39	LY	0.3579	66.9800	0.2800	65.3800	62.0500
40	HG Market Research	0.3272	68.3000	0.0000	61.0600	61.7100
41	qingqiu-test	0.3442	66.7500	0.3300	63.8000	61.7000
42	rw	0.3473	67.4900	0.0000	63.9300	61.6700
43	MaxwellEQ (aside_v1)	0.3458	67.3200	0.0000	62.6100	61.5200
44	try my best	0.3335	67.8900	0.0000	60.5700	61.1600
45	OP	0.3530	55.2800	0.9300	62.2900	60.6500
46	Tzachristas_team	0.3606	46.8800	0.2800	62.6900	60.1900
47	dunge	0.3503	66.9200	0.0400	60.0300	60.1100
48	ltrc	0.3642	47.9600	0.1900	60.9400	59.9000
49	Capy Awesome	0.3771	52.4000	1.4400	58.9800	58.4800
50	Intelligence System Lab	0.4104	66.7700	0.1900	60.4800	57.5400
51	BoundedAccord (Tesseract 5 Layout Baseline)	0.4996	-0.7500	0.1000	46.5600	44.9000
52	Frontier	0.5997	27.0900	0.0000	44.6600	39.5300
53	DocBench AI	0.6213	1.4100	0.1400	46.0300	38.9200
54	InvestigationBridge	0.6073	13.5900	0.2600	40.1000	36.8900

Sorted by: Overall · Official rank remains based on Overall

Select a metric heading to compare systems. Lower is better for Text ED; higher is better for all other metrics.

Dataset

Challenge Data

The public Dr.DocBench dataset and annotations are available on Hugging Face for development and format reference.

Open on Hugging Face

Competition Evaluation

Official evaluation inputs, sample submissions, and validation tools are provided through the Dr.DocBench EvalAI challenge.

Open EvalAI Challenge

Evaluation

Submissions are evaluated using four component metrics plus one Overall aggregate. Final rankings are determined by the Overall score.

Text Edit Distance

Normalized Levenshtein distance over scored text categories, from 0–1. Lower is better.

Table TEDS

Mean tree-edit-distance similarity for valid tables, shown on a 0–100 scale. Higher is better.

Formula CDM

Mean CDM F1 for isolated display formulas, shown on a 0–100 scale. Higher is better.

Reading Order

(1 − normalized reading-order edit distance) × 100. Higher is better.

Overall

Available component scores are converted to a 0–100 scale and averaged for each scorable page, then averaged across all scorable pages. Missing components are excluded from the page denominator rather than scored as zero; non-scorable pages are excluded.

Submission

1. Prepare

Download the relevant evaluation package and generate one prediction for every released page identifier. The challenge uses a single-page prediction unit, unlike the default two-page window setting described in the Dr.DocBench paper.

submission.zip
└── predictions.jsonl
    └── required fields: subject, document_id, page, markdown

The evaluator applies strict checks for missing, extra, or duplicate predictions and invalid identifiers or submission structure.

A canonical mds/ Markdown tree mirroring the released hierarchy is also accepted.

2. Submit

Upload the validated ZIP through EvalAI. Final rankings will be determined through the Private Test phase.

Open EvalAI Challenge

Rules

Eligibility

Open to researchers and teams worldwide. There is no team-size limit.

Submissions

Maximum three submissions per day. Each team may select up to two submissions for final evaluation.

Requirements

Top-ranked teams must submit a technical report. All external data, models, and resources must be disclosed.