Leaderboard

Current

Live standings from the Kaggle public leaderboards

Overview Task Timeline Rules Participate Leaderboard Publications Organisers FAQ

BAcc@10 over time

Each dot is a team improving its own best public BAcc@10 score; the stepped line follows the best score overall. Hover a dot for the team.

team improves its own bestnew best overallstate of the art

0.719

Best BalAcc@10

Teams

1,271

Submissions

—

Days remaining

Metric:

Teams by BalAcc@10

Top-10 balanced accuracy on the 50-word vocab — Kaggle’s official public score

1Varshith Madishetty

0.7190

2titus fisher

0.7173

3Umur Yıldız

0.7148

4MindLink

0.7112

5ottietjesesakfs

0.7104

6Infera-Neuro

0.6786

7Kim

0.6752

8Munich Logic Lab

0.6729

9Jordan Griffith

0.6694

10CWBZ

0.6601

11dav0dea

0.6563

12JovanaLab

0.6558

13MagnetoLex

0.6548

14kakuteki

0.6463

15neural2speech

0.6447

16Michal

0.6438

17@Arnauya

0.6428

18Jatin Arutla

0.6399

19Brain King

0.6362

20SENPAI

0.6296

Showing top 20 of 40 teams — full standings in the table below.

Best score per team — every metric

Kaggle ranks one number. We archive every submission and re-score it ourselves, so each team's best on each metric sits side by side (public split). Click a metric to rank by it.

#	Team	BalAcc@10 ↓	BalAcc@1	OVMI	Subs
1	Varshith Madishetty	0.7190	0.2887	1.1267	55
2	titus fisher	0.7173	0.2680	1.1175	99
3	Umur Yıldız	0.7148	0.2432	1.1690	110
4	MindLink	0.7112	0.2539	1.1265	118
5	ottietjesesakfs	0.7104	0.2587	1.1408	100
6	Infera-Neuro	0.6786	0.2157	1.1512	52
7	Kim	0.6752	0.2124	1.0237	79
8	Munich Logic Lab	0.6729	0.1997	1.0832	35
9	Jordan Griffith	0.6694	0.2386	1.1628	28
10	CWBZ	0.6601	0.2089	1.1239	33
11	dav0dea	0.6563	0.2223	1.1486	13
12	JovanaLab	0.6558	0.2729	1.0757	26
13	MagnetoLex	0.6548	0.2173	1.1227	39
14	kakuteki	0.6463	0.2239	1.1474	47
15	neural2speech	0.6447	0.1447	1.1265	10
16	Michal	0.6438	0.1418	1.0159	3
17	@Arnauya	0.6428	0.1899	1.1423	81
18	Jatin Arutla	0.6399	0.2387	1.0920	12
19	Brain King	0.6362	0.1522	1.1299	55
20	SENPAI	0.6296	0.1858	1.0606	61
21	onlytry	0.6152	0.1934	1.0748	9
22	Artem Blanar	0.5891	0.1831	1.0056	29
23	Connor Finnerty	0.5191	0.1227	1.0028	22
24	syouya tobita	0.5036	0.1062	1.0743	5
25	Baseline	0.4444	0.0733	0.9803	1
26	Sia	0.4362	0.1014	0.9362	5
27	Shuntaro Suzuki	0.3540	0.0488	0.8104	5
28	MindReaders2.0	0.3322	0.0740	0.9441	78
29	Jun Jie Li	0.3176	0.0864	0.8436	12
30	peperonata	0.3066	0.0323	0.7890	3
31	monte-carlo	0.2900	0.0483	0.4671	5
32	Bhargav Kowshik	0.2871	0.0292	1.0947	16
33	AutoDecode	0.2602	0.0405	0.9752	10
34	Fon1as	0.2538	0.0249	0.2581	1
35	willguido	0.2444	0.0244	0.1239	4
36	Nguyen Anh	0.2164	0.0311	0.5181	3
37	Salsinats	0.2000	0.0200	0.4997	2
38	lavender_lover	0.1906	0.0200	0.4893	3
39	LizaSheina	0.1880	0.0284	0.3681	1
40	Viknesh .V	0.1600	0.0200	0.0000	1

About these numbers

BalAcc@10 is Kaggle's official public score, shown here exactly as Kaggle reports it. BalAcc@1 (strict top-1) and OVMI (the information a decoder recovers over the vocabulary) are computed by us from the submitted files, on the same public split — final placements use the private split, revealed after the deadline.

Updated … · Deep Track on Kaggle →