Share your thoughts, 1 month free Claude Pro on usSee more

Image Classification on 8 Vision Tasks (test)

94.6Avg Accuracy

Experts

Updated 15d ago

Evaluation Results

Method
Experts 2026.04	94.6	-	-	-	-	-	-	-	-	-
theta_B fine-tune 2026.02	94.59	-	-	-	-	-	-	-	-	20.96
Fine-tuned 2025.05	94.2	-	-	-	-	-	-	-	-	-
Experts 2026.04	94.1	-	-	-	-	-	-	-	-	-
MDA TA 2025.11	93.6	78.8	90.7	94.3	99.2	95.2	97	99.6	94.5	-
Multitask 2025.05	93.5	-	-	-	-	-	-	-	-	-
ISO TA 2025.11	93.3	79.9	91.4	94.8	99	90.5	95.5	99.3	96.3	-
θB 2026.02	93.23	-	-	-	-	-	-	-	-	25.83
theta_A fine-tune 2026.02	93.23	-	-	-	-	-	-	-	-	19.6
theta_A fine-tune 2026.02	92.96	-	-	-	-	-	-	-	-	34.2
TSV TA 2025.11	92.9	77.8	89.8	93.5	98.7	94.7	96.1	99.5	93.2	-
ISO-CLS TA 2025.11	92.9	79.6	91.7	94.6	98.8	88.8	95.4	99.2	95.6	-
ACTMAT 2026.04	92.2	-	-	-	-	-	-	-	-	-
TSV 2026.04	91.1	-	-	-	-	-	-	-	-	-
Fine-tuned 2025.05	90.5	-	-	-	-	-	-	-	-	-
ISO-C 2026.04	90.5	-	-	-	-	-	-	-	-	-
Experts 2026.04	90.4	-	-	-	-	-	-	-	-	-
theta_B fine-tune 2026.02	90.39	-	-	-	-	-	-	-	-	31.63
REGMEAN 2026.04	90	-	-	-	-	-	-	-	-	-
ACTMAT 2026.04	89.5	-	-	-	-	-	-	-	-	-
Multitask 2025.05	88.9	-	-	-	-	-	-	-	-	-
TSV 2026.04	88.8	-	-	-	-	-	-	-	-	-
ISO-C 2026.04	88.4	-	-	-	-	-	-	-	-	-
Individual 2026.03	87.82	72.49	76.61	93.97	98.19	96.5	98.29	99.15	67.34	-
NPS 2025.05	87.6	-	-	-	-	-	-	-	-	-
REGMEAN 2026.04	87.6	-	-	-	-	-	-	-	-	-
MDA TA 2025.11	87.4	70.3	72	87.9	96	92.5	93.3	99.3	87.5	-
Consensus TIES 2025.05	86.2	-	-	-	-	-	-	-	-	-
Ties-Merging 2025.05	86	-	-	-	-	-	-	-	-	-
Consensus TA 2025.05	85.8	-	-	-	-	-	-	-	-	-
TSV TA 2025.11	85.7	67.2	70.3	85.7	94.5	91.6	92.2	99.3	84.8	-
ISO TA 2025.11	84.9	71.8	73.8	87.7	90.9	82.3	88.1	98.2	86	-
Task Arithmetic 2025.05	84.5	-	-	-	-	-	-	-	-	-
TA 2026.04	84.3	-	-	-	-	-	-	-	-	-
ISO-CLS TA 2025.11	83.8	72.4	74	87.9	90.6	76.9	85.3	97.5	85.5	-
RegMean 2025.05	83.7	-	-	-	-	-	-	-	-	-
TSV 2026.04	83.3	-	-	-	-	-	-	-	-	-
REGMEAN 2026.04	83	-	-	-	-	-	-	-	-	-
ACTMAT 2026.04	82.9	-	-	-	-	-	-	-	-	-
Fisher Merging 2025.05	82.2	-	-	-	-	-	-	-	-	-
ISO-C 2026.04	81.6	-	-	-	-	-	-	-	-	-
THESEUS 2026.02	80.24	-	-	-	-	-	-	-	-	6.61
Averaging 2025.05	79.6	-	-	-	-	-	-	-	-	-
AVERAGE 2026.04	79.3	-	-	-	-	-	-	-	-	-
THESEUS 2026.02	77.32	-	-	-	-	-	-	-	-	9.92
NPS 2025.05	76.5	-	-	-	-	-	-	-	-	-
TA 2026.04	76.1	-	-	-	-	-	-	-	-	-
THESEUS 2026.02	76.02	-	-	-	-	-	-	-	-	8.62
THESEUS 2026.02	75.58	-	-	-	-	-	-	-	-	8.18
THESEUS 2026.02	73.89	-	-	-	-	-	-	-	-	6.49
theta_B zero-shot 2026.02	73.63	-	-	-	-	-	-	-	-	0
Ties-Merging 2025.05	73.6	-	-	-	-	-	-	-	-	-
Consensus TA 2025.05	73.5	-	-	-	-	-	-	-	-	-
Consensus TIES 2025.05	73.3	-	-	-	-	-	-	-	-	-
THESEUS 2026.02	72.5	-	-	-	-	-	-	-	-	5.1
AVERAGE 2026.04	72.2	-	-	-	-	-	-	-	-	-
RegMean 2025.05	71.8	-	-	-	-	-	-	-	-	-
Task Arithmetic 2025.05	70.1	-	-	-	-	-	-	-	-	-
TA 2026.04	69.7	-	-	-	-	-	-	-	-	-
THESEUS 2026.02	69	-	-	-	-	-	-	-	-	10.24
Fisher Merging 2025.05	68.3	-	-	-	-	-	-	-	-	-
THESEUS 2026.02	68.09	-	-	-	-	-	-	-	-	9.33
THESEUS 2026.02	67.62	-	-	-	-	-	-	-	-	8.86
θB 2026.02	67.4	-	-	-	-	-	-	-	-	0
THESEUS 2026.02	65.9	-	-	-	-	-	-	-	-	7.14
Averaging 2025.05	65.8	-	-	-	-	-	-	-	-	-
AVERAGE 2026.04	65.5	-	-	-	-	-	-	-	-	-
Zeroshot 2026.04	65.2	-	-	-	-	-	-	-	-	-
THESEUS 2026.02	64.52	-	-	-	-	-	-	-	-	5.76
DC-Merge 2026.03	64.17	64.98	62.93	75.29	61.15	56.93	57.27	84.79	50	-
Iso-CTS 2026.03	63.01	63.68	60.72	72.87	64.26	55.1	52.31	82.8	52.34	-
TSV-M 2026.03	58.91	63.84	62.66	69.37	55.44	50.66	46.88	75.86	46.54	-
theta_B zero-shot 2026.02	58.76	-	-	-	-	-	-	-	-	0
KnOTS-TIES 2026.03	55.93	63.48	61.39	65.49	48.26	49.36	43.42	71.24	44.79	-
Zeroshot 2026.04	55.5	-	-	-	-	-	-	-	-	-
WUDI-Merging 2026.03	55.25	63.32	60.89	67.4	51.82	46.12	39.9	66.96	45.64	-
Task Arithmetic 2026.03	52.8	62.61	59.82	65.78	46.52	45.25	34.71	63.43	44.31	-
Zeroshot 2026.04	48.2	-	-	-	-	-	-	-	-	-
Zeroshot 2026.03	47.88	63.21	59.49	60.25	44.3	31.42	32.27	47.94	44.15	-
TA 2026.04	41.8	-	-	-	-	-	-	-	-	-
TA 2026.04	26.5	-	-	-	-	-	-	-	-	-
TA 2026.04	24.3	-	-	-	-	-	-	-	-	-