All Projects → CLUEbenchmark → MobileQA

CLUEbenchmark / MobileQA

Licence: other
离线端阅读理解应用 QA for mobile, Android & iPhone

Programming Languages

python
139335 projects - #7 most used programming language
java
68154 projects - #9 most used programming language

Projects that are alternatives of or similar to MobileQA

Cluedatasetsearch
搜索所有中文NLP数据集,附常用英文NLP数据集
Stars: ✭ 2,112 (+4210.2%)
Mutual labels:  qa, chinese, machine-reading-comprehension
Clue
中文语言理解测评基准 Chinese Language Understanding Evaluation Benchmark: datasets, baselines, pre-trained models, corpus and leaderboard
Stars: ✭ 2,425 (+4848.98%)
Mutual labels:  chinese, albert, bert
CLUE pytorch
CLUE baseline pytorch CLUE的pytorch版本基线
Stars: ✭ 72 (+46.94%)
Mutual labels:  chinese, albert, bert
BERT-chinese-text-classification-pytorch
This repo contains a PyTorch implementation of a pretrained BERT model for text classification.
Stars: ✭ 92 (+87.76%)
Mutual labels:  chinese, bert
AiSpace
AiSpace: Better practices for deep learning model development and deployment For Tensorflow 2.0
Stars: ✭ 28 (-42.86%)
Mutual labels:  chinese, bert
NLPDataAugmentation
Chinese NLP Data Augmentation, BERT Contextual Augmentation
Stars: ✭ 94 (+91.84%)
Mutual labels:  chinese, bert
Nlp chinese corpus
大规模中文自然语言处理语料 Large Scale Chinese Corpus for NLP
Stars: ✭ 6,656 (+13483.67%)
Mutual labels:  chinese, bert
OpenDialog
An Open-Source Package for Chinese Open-domain Conversational Chatbot (中文闲聊对话系统,一键部署微信闲聊机器人)
Stars: ✭ 94 (+91.84%)
Mutual labels:  chinese, bert
Transformer-QG-on-SQuAD
Implement Question Generator with SOTA pre-trained Language Models (RoBERTa, BERT, GPT, BART, T5, etc.)
Stars: ✭ 28 (-42.86%)
Mutual labels:  albert, bert
bert nli
A Natural Language Inference (NLI) model based on Transformers (BERT and ALBERT)
Stars: ✭ 97 (+97.96%)
Mutual labels:  albert, bert
tfbert
基于tensorflow1.x的预训练模型调用,支持单机多卡、梯度累积,XLA加速,混合精度。可灵活训练、验证、预测。
Stars: ✭ 54 (+10.2%)
Mutual labels:  albert, bert
classifier multi label
multi-label,classifier,text classification,多标签文本分类,文本分类,BERT,ALBERT,multi-label-classification
Stars: ✭ 127 (+159.18%)
Mutual labels:  albert, bert
MRC Competition Dureader
机器阅读理解 冠军/亚军代码及中文预训练MRC模型
Stars: ✭ 552 (+1026.53%)
Mutual labels:  qa, bert
embedding study
中文预训练模型生成字向量学习,测试BERT,ELMO的中文效果
Stars: ✭ 94 (+91.84%)
Mutual labels:  chinese, bert
Roberta zh
RoBERTa中文预训练模型: RoBERTa for Chinese
Stars: ✭ 1,953 (+3885.71%)
Mutual labels:  chinese, bert
NLP-paper
🎨 🎨NLP 自然语言处理教程 🎨🎨 https://dataxujing.github.io/NLP-paper/
Stars: ✭ 23 (-53.06%)
Mutual labels:  albert, bert
ALBERT-Pytorch
Pytorch Implementation of ALBERT(A Lite BERT for Self-supervised Learning of Language Representations)
Stars: ✭ 214 (+336.73%)
Mutual labels:  albert, bert
FewCLUE
FewCLUE 小样本学习测评基准,中文版
Stars: ✭ 251 (+412.24%)
Mutual labels:  chinese, bert
classifier multi label seq2seq attention
multi-label,classifier,text classification,多标签文本分类,文本分类,BERT,ALBERT,multi-label-classification,seq2seq,attention,beam search
Stars: ✭ 26 (-46.94%)
Mutual labels:  albert, bert
Medi-CoQA
Conversational Question Answering on Clinical Text
Stars: ✭ 22 (-55.1%)
Mutual labels:  albert, bert

MobileQA

TensorFlow Lite BERT QA Android Example Applicationtflite-android-transformers 展示了基于Bert/DistilBERT的离线QA例子,但是只支持英文和安卓设备。

本项目计划实现基于中文的机器阅读理解在手机端的离线应用,并且同时支持安卓和苹果设备。

Targeting to release before Dec 5th. 目标是12月5日前发布。

数据集

使用CMRC 2018 公开数据集,该数据集是第二届讯飞杯中文机器阅读理解评测所使用的数据。数据集已被计算语言学顶级国际会议EMNLP 2019录用

模型

使用 albert_zh_small 预训练模型,额外加上一层全连接做answer span预测。

  • 在CMRC2018数据集的验证集上,max_seq_len为512的模型得分为F1:75.989, EM:52.097, Average:64.038,max_seq_len为384的模型得分为F1:74.781, EM:51.010, Average:62.895

  • max_seq_len为512的模型使用tflite转换后大小为18M,经测试,该模型在4线程的安卓手机上推理延时为580ms左右,在单线程条件下为1.4s左右。

  • max_seq_len为384的模型使用tflite转换后大小为18M,经测试,该模型在4线程的安卓手机上推理延时为390ms左右,在单线程条件下为930ms左右

  • 模型训练与模型转换过程见bert_cn_finetune-master

Android Demo

已完成第一版,详见 tflite-android-transformers-master

效果示例:

android demo

IOS Demo

进行中

Updates

Contribution

如果你感兴趣或希望提供帮助,发送邮件[email protected]

If you're interested or want to help, send an email to [email protected]

Note that the project description data, including the texts, logos, images, and/or trademarks, for each open source project belongs to its rightful owner. If you wish to add or remove any projects, please contact us at [email protected].