Issues

• 关于分词示例roberta_participle_CRF_train.py

  920232796/bert_seq2seq#71 · Filter by author Turskey opened on Nov 20, 2023

• 数据集分享

  920232796/bert_seq2seq#70 · Filter by author zhangzai666 opened on May 12, 2023

• 我从新改变了vocab.txt的大小，在加载预训练模型的时候出现维度不匹配，请问应该怎么解决了？我发现代码里面有个keep_tokens,似乎可以解决，但我不知道keep_tokens改如何设置。

  920232796/bert_seq2seq#68 · Filter by author panbo-bridge opened on Mar 22, 2023

• 多卡训练失败，目前有4个gpu，但使用的时候一直显示用的cuda0，指定其他cuda也不生效

  920232796/bert_seq2seq#66 · Filter by author GUORUIWANG opened on Nov 30, 2022

• 关于UniLM

  920232796/bert_seq2seq#65 · Filter by author cppww opened on Oct 24, 2022
  1

• 关于细粒度NER数据集

  920232796/bert_seq2seq#64 · Filter by author the9527001 opened on Oct 8, 2022

• 自動摘要 fine tune

  920232796/bert_seq2seq#63 · Filter by author frankShih opened on Aug 30, 2022
  1

• 您好 请问在simbert模型里面 seq2seq的loss比similarity的loss大的多是正常的吗？

  920232796/bert_seq2seq#61 · Filter by author EddieChen324 opened on Jul 10, 2022
  3

• 你好，我想问一下，关于gpt文章写作的数据集应该是什么样的格式，是一小段标题作为输入，文章作为输出么？

  920232796/bert_seq2seq#58 · Filter by author viego1999 opened on May 15, 2022
  5

• 请问在做标题生成的时候支持point_net对应的copy机制吗？

  920232796/bert_seq2seq#57 · Filter by author chenzelong opened on Apr 18, 2022
  1

• 如何能联系到您呢？qq群加不进去，有关于epoch和bart做seq2seq方面的问题想请教一下> > 您好 在不断降低学习率之后loss还是卡在原来的位置，请问如何能与您取得联系？

  920232796/bert_seq2seq#56 · Filter by author chenzelong opened on Apr 15, 2022
  5

• unilm多少epoch收敛比较好？

  920232796/bert_seq2seq#55 · Filter by author chenzelong opened on Apr 15, 2022
  1

• 请问bart做标题生成怎么跑起来？

  920232796/bert_seq2seq#54 · Filter by author chenzelong opened on Apr 15, 2022
  4

• 您好，我想添加一些我自己的数据集做fine-tune，请问应该怎么做呢

  920232796/bert_seq2seq#52 · Filter by author pangsg opened on Mar 14, 2022
  1

• roberta_auto_title_train.py

  920232796/bert_seq2seq#50 · Filter by author lishuzhen97 opened on Feb 11, 2022
  4

• 1. 关于多卡训练支持建议。2. 关于tokenizer的pre_train建议。3. 关于外露模型配置API的建议。

  920232796/bert_seq2seq#49 · Filter by author PolarisRisingWar opened on Dec 10, 2021
  1

• simbert_model.py文件中的117行

  920232796/bert_seq2seq#48 · Filter by author singaln opened on Dec 7, 2021
  5

• 模型加载输出问题

  920232796/bert_seq2seq#47 · Filter by author yang41690104 opened on Dec 1, 2021
  5

• 如何使用问题？

  920232796/bert_seq2seq#46 · Filter by author Jockz opened on Nov 25, 2021
  1

• 关于数据集问题

  920232796/bert_seq2seq#45 · Filter by author XingyuZhu-Pamela opened on Nov 17, 2021
  2

• 关于医学代码的问题

  920232796/bert_seq2seq#43 · Filter by author cc846049716 opened on Oct 7, 2021
  16

• 三元组信息抽取中随机抽取object，最终只用到了object开始、结束字符对应的编码用于predict、subject预测，这样是否合理呢

  920232796/bert_seq2seq#42 · Filter by author jackjqli opened on Sep 15, 2021
  1

• 粗粒度NER-tokenizer问题

  920232796/bert_seq2seq#41 · Filter by author NiceMartin opened on Aug 1, 2021
  1

• 代码问题 multi-head attention 中的 extended_attention_mask

  920232796/bert_seq2seq#39 · Filter by author NiceMartin opened on Jul 17, 2021
  4

• 关于标题分类的问题

  920232796/bert_seq2seq#38 · Filter by author Evialzhang opened on Jul 15, 2021
  1