CLUE1.0分类任务排行榜     CLUE1.1/1.0提交规则  |   项目地址
CLUE1.1与CLUE1.0区别:区别与原有的CLUE1.0,CLUE1.1在部分任务启用了新的测试集,训练集和验证集保持不变;CLUE1.0保留CMNLI自然语言推理任务

模型
排行模型研究机构测评时间Score1.0认证AFQMCTNEWS1.0IFLYTEKCMNLIOCNLI_50KWSC1.0CSL
1TI-NLP优图实验室 & 腾讯云21-10-1983.251待认证82.779.365.2384.3184.5796.5590.1
2ShenZhouQQ浏览器实验室(QQ Browser Lab)21-09-1983.247待认证80.5574.1567.6586.4986.3796.5590.97
3HUMANCLUE19-12-0182.943已认证817180.37690.39884
4Mengzi澜舟科技-创新工场21-09-1482.436待认证81.7975.0665.0886.1382.5796.5589.87
5BERTSGSogou Search21-06-2581.991待认证79.8574.1564.5485.385.9395.1789
6MotianQQ浏览器搜索21-06-2581.764待认证78.373.1865.4685.4484.9794.8390.17
7Pangu华为云-循环智能21-04-2381.016待认证78.1172.0765.1985.1983.395.5287.73
8PLUGAlibaba DAMO NLP21-04-1880.614待认证77.4473.066484.9583.2794.4887.1
9Bertlihaiyu21-04-0879.663待认证75.670.3264.9284.5581.7393.4587.07
10MT-BERTsMeituan NLP21-03-1079.624待认证77.3670.0364.3185.1483.4789.6687.4
11Knowledge-based姜汁柠檬21-07-2379.611待认证76.8770.263.7387.9781.1392.4184.97
12LICHEE腾讯看点21-01-0879.364待认证76.9770.564.1584.5481.390.6987.4
13roberta_selfrunOPPO小布助手21-09-2979.269待认证77.8869.3763.9282.9480.493.187.27
14UER-ensembleTencentPretrain & TI-ONE20-11-2879.154待认证76.8272.26484.0980.890.3485.83
15BERTsBERTs20-12-2479.107待认证76.7769.9463.9284.4882.988.9786.77
16Archer-24E-SINGLEsearch-nlp20-12-2479.086待认证77.2669.5462.2785.2383.579085.73
17selfrun-ensembleOPPO小布助手20-12-2278.674待认证76.0969.163.9282.5680.491.3887.27
18dfasdfadfadfafdaf22-10-1278.661待认证76.7268.3163.3184.9881.188.2887.93
19roformer&erlangshenhuangjh22-11-0978.661待认证76.7268.3163.3184.9881.188.2887.93
20Archer-24lsearch-nlp20-11-3078.550待认证77.4469.9662.6984.7882.5787.2485.17

ALBERT(Ensemble)

GitHub/模型网址:

提交日期:9月17日

分数:9月17日

更多详情:

型号说明

阿尔伯特模型集合

参数说明

单任务微调。我们从MNLI为RTE、STS和MRPC优化的模型开始

总参数:-1

共享参数:-1

诊断信息

诊断主混淆矩阵

C N E
C 182 36 40
N 81 189 116
E 17 69 374

C = 对立

N = 不包含

E = 包含

类别相关马修分数

获取排行榜数据成功!