ymcui/Chinese-ELECTRA

Issues

• max length of hfl/chinese-electra-180g-large-discriminator tokenizer is incorrect

  #93 opened a year ago by AngledLuffa
  0

• Pytorch版本

  #92 opened a year ago by Freshman-Fan
  1

• 关于微调后的模型

  #91 opened a year ago by Yyy11181
  2

• 预训练数据文件格式，是一行一个句子，还是一行一个包含多个句子的文档

  #90 opened a year ago by hitwangshuai
  11

• huggingface上的hfl/chinese-legal-electra-base-discriminator缺失vocab.txt文件

  #87 opened a year ago by JasonCZH4
  6

• 百度网盘里的ELECTRA-180g-base, Chinese 包括generator部分的参数吗

  #89 opened 2 years ago by hitwangshuai
  2

• 微调后的模型大小问题

  #88 opened 2 years ago by zidong-onepiece1
  2

• 应用electra-discriminator训练过的模型作分类任务时部分权重没用上

  #84 opened 2 years ago by TinykWong
  3

• ELECTRA的Generator为什么没有选择修改为WWM？

  #83 opened 2 years ago by johnsongwx
  2

• 预训练语料的疑问

  #82 opened 2 years ago by Rommle
  3

• 一个原理上的疑问

  #81 opened 2 years ago by Ricardokevins
  2

• Some weights of the model checkpoint at hfl/chinese-electra-small-ex-generator were not used when initializing

  #63 opened 4 years ago by SunYanCN
  6

• 请教司法领域版本的预训练细节

  #78 opened 3 years ago by ken-cina
  2

• 参数中的discriminator_predictions是什么？

  #76 opened 3 years ago by lotress
  2

• Bug in preprocessing data?

  #77 opened 3 years ago by mrzrx
  2

• Huggingface 的 legal base 和 legal small tf版是同一个模型

  #75 opened 3 years ago by OpenWaygate
  1

• 关于并行

  #40 opened 4 years ago by Hanlard
  5

• Key discriminator_predictions/dense/bias/adam_m not found in checkpoint

  #74 opened 4 years ago by wzqy56789
  2

• finetune/qa_tasks.py的585行有个错误

  #73 opened 4 years ago by rancho628
  2

• 用electra-small在CMRC上微调时，训练过程没有问题，但是测试时出现了写出结果错误

  #72 opened 4 years ago by Rwkun
  2

• 关于实验所需的显卡配置

  #71 opened 4 years ago by rxc205
  0

• 请问训练之后转换为pb文件，output的名称是什么？

  #69 opened 4 years ago by jimzeus
  2

• huggingface.co上chinese-electra-180g-small-discriminator的文件里缺少tokenizer.json

  #70 opened 4 years ago by zihoudanyuan
  5

• 请问可以直接使用Paddle Hub或者Hugging Face上的chinese-electra-small模型进行情感分析任务吗？

  #68 opened 4 years ago by jimzeus
  3

• 关于复现LCQMC任务的一些疑问

  #67 opened 4 years ago by GJHHHH
  6

• 预训练时disc_recall disc_precision 都是0

  #66 opened 4 years ago by ZJiaBin
  3

• 关于在领域数据集上继续预训练后微调效果很差的问题

  #38 opened 4 years ago by sc1054
  10

• 请问下这个项目和google electra 项目的区别

  #62 opened 4 years ago by 652994331
  6

• 判别器的结构

  #65 opened 4 years ago by Azuk1
  1

• 模型包括判别器bert输出后的两层全连接神经网络权重吗？

  #61 opened 4 years ago by zejunwang1
  4

• Bert service 使用 chinese-electra-base 模型，输出结果有问题

  #58 opened 4 years ago by Echo0117
  4

• 大语料版预计什么时候上 paddlehub？

  #59 opened 4 years ago by T-baby
  2

• 在下游中文ner任务中, electra各个模型学习率应该设置多大？

  #60 opened 4 years ago
  1

• 请问中文的预训练，mask的粒度是字符粒度，这有什么道理吗，为什么不采用词粒度呢

  #53 opened 4 years ago by meteorshowers
  8

• 关于预训练是否加入了 wwm

  #56 opened 4 years ago by 652994331
  3

• 请告知params_cmrc2018.json这个文件所在之处

  #44 opened 4 years ago by ChrisChaw
  2

• 无法加载分词器以及模型

  #55 opened 4 years ago by zhangatao
  0

• 关于训练数据

  #54 opened 4 years ago by zshy1205
  1

• Sentiment classification

  #52 opened 4 years ago by ML-PPAP
  5

• keyError:3200000

  #50 opened 4 years ago by wanqiufenglv
  3

• download issue

  #51 opened 4 years ago by ML-PPAP
  3

• error in loading checkpoints for pretraining

  #49 opened 4 years ago by w5688414
  3

• 请问finetune时应如何设置token type id?

  #48 opened 4 years ago by AOZMH
  4

• ELECTRA-base的配置文件的参数设置

  #46 opened 4 years ago by ChrisChaw
  21

• electra base的check point file在哪里？

  #47 opened 4 years ago by ChrisChaw
  1

• 询问项目入口文件与项目启动命令

  #45 opened 4 years ago by ChrisChaw
  1

• how to download a tf2.0 model of electra-hongkongese-large-discriminator ??

  #42 opened 4 years ago by marcusau
  2

• 预训练数据量

  #41 opened 4 years ago by frostjsy
  2

• 模型加载兼容问题

  #39 opened 4 years ago by frostjsy
  3

• 关于预训练，步骤可以写的详细一点吗？

  #37 opened 4 years ago by Hanlard
  1