Codesota · Models · DeBERTa-v3-largeMicrosoft6 results · 5 benchmarks
Model card

DeBERTa-v3-large.

Microsoftopen-source304M paramsDeBERTa-v3-large

DeBERTaV3. ICLR 2023. GLUE average 91.37.

§ 01 · Benchmarks

Every benchmark DeBERTa-v3-large has a recorded score for.

#BenchmarkArea · TaskMetricValueRankDateSource
01GLUENatural Language Processing · Fill-Maskavg-score91.4%#1/32023-01-01source ↗
02SQuAD v2.0Natural Language Processing · Question Answeringem88.4%#1/22021-11-18source ↗
03SQuAD v2.0Natural Language Processing · Question Answeringf191.4%#1/222021-11-18source ↗
04SuperGLUENatural Language Processing · Text Classificationaverage-score91.4%#1/72021-11-18source ↗
05CoNLL-2003Natural Language Processing · Named Entity Recognitionf193.4%#2/72021-11-18source ↗
06SNLINatural Language Processing · Natural Language Inferenceaccuracy92.2%#2/82021-11-18source ↗
Rank column shows this model’s position vs all other models scored on the same benchmark + metric (competitors after the slash). #1 in red means current SOTA. Sorted by rank, then newest result.
§ 02 · Strengths by area

Where DeBERTa-v3-large actually performs.

Natural Language Processing
5
benchmarks
avg rank #1.3
§ 03 · Papers

2 papers with results for DeBERTa-v3-large.

  1. 2023-01-01· Natural Language Processing· 1 result

    DeBERTaV3: Improving DeBERTa using ELECTRA-Style Pre-Training with Gradient-Disentangled Embedding Sharing

  2. 2021-11-18· Natural Language Processing· 5 results

    DeBERTaV3: Improving DeBERTa using ELECTRA-Style Pre-Training with Gradient-Disentangled Embedding Sharing

§ 04 · Related models

Other Microsoft models scored on Codesota.

RAD-DINO
2 results · 1 SOTA
NaturalSpeech 3
~500M params · 1 result · 1 SOTA
Swin Transformer V2 Large
197M params · 1 result · 1 SOTA
WavLM Large (SV)
316M params · 1 result · 1 SOTA
ResNet-50
25M params · 3 results
Florence-2-Large
2 results
KOSMOS-2.5
2 results
ResNet-152
60M params · 2 results
§ 05 · Sources & freshness

Where these numbers come from.

arxiv
6
results
6 of 6 rows marked verified. · first result 2021-11-18, latest 2023-01-01.