Training a model

[salbatarni]

Hello, I have been trying to replicate the training for you model, however I have not quite succeded so I want to make sure of some things

• bert tokenizer used is bert-base-uncased?
• in the training the pre-trained model passed to DocumentBertCombineWordDocumentLinear and DocumentBertSentenceChunkAttentionLSTM is also bert-base-uncased, with the first 11 layers frozen correct?

[iamhere1]

Hello, I have been trying to replicate the training for you model, however I have not quite succeded so I want to make sure of some things

• bert tokenizer used is bert-base-uncased?
• in the training the pre-trained model passed to DocumentBertCombineWordDocumentLinear and DocumentBertSentenceChunkAttentionLSTM is also bert-base-uncased, with the first 11 layers frozen correct?

1. Bert tokenizer used is bert-base-uncased
2. Yes, all the pre-trained models used are bert-base-uncased, and finetuned with the first 11 layers frozen

[shield124]

Hello, I have been trying to replicate the training for you model, however I have not quite succeded so I want to make sure of some things

• bert tokenizer used is bert-base-uncased?
• in the training the pre-trained model passed to DocumentBertCombineWordDocumentLinear and DocumentBertSentenceChunkAttentionLSTM is also bert-base-uncased, with the first 11 layers frozen correct?

请问你最后有实现训练模型的代码吗？可以分享下吗？

[shield124]

Could you please provide the code you used to train the model? Can you share it with me?