githublsk

encounting the below error, please help me

2

when i covert pytoch model to onnx, the error occur as below, I think because entities and relations is a dict format data, which I set them with wrong shape?...

custom op torch.nn.bilinear to export for onnx(so urgen, Great thanks)

3

when I convert pytorch model, the torch.nn.bilinear op can not be exported, I just add the blew code using custom op, the model can be convert successfully and onnx check...

Hi @ChaseMonsterAway , @mileistone I have one question that how can I train the model use multi gpu on one machine, can you help us to solve the question? Thank...

Torch.nn.bilinear can not export to onnx

1

Hi developer, When I export layoutxlm RE model to onnx, I find torch.nn.bilinear can not export to onnx, can a function that can be supported by onnx replace torch.nn.bilinear? 

I use layoutxlm to training my data for downstream, when I convert the model which I trained to onnx by huggingface code layoutlmv2-to-onnx, it occurs below problem, can you give...

some questions about SER and RE task inference

2

Hi [wolfshow](https://github.com/wolfshow), I have a question on the belowing: I using Chinese document for layoutxlm downstream RE task following below steps: step1: finetuning a SER model with Chinese document; step2:finetuning...

Is layloutlmv3 support for SER and RE work for xfund dataset?

4

Model description: LayoutLMV3 is the successor of the LayoutLM models. I have a question that dose layoutlmv3 support for SER and RE work for xfund dataset?

SER和RE联合训练

2

开发者您好， 我们最近正在用layoutxlm做信息抽取任务，效果还是不错的，但是目前的问题是需要单独训练SER和RE两个模型，这两个模型目前看还是比较重的，目前想把这两个模型整合到一起，目前调研主流的观点是：先NER（命名实体识别）再RE（关系抽取），实体和关系共享同一个网络编码，比如pretrained_model的基础上，接一个NER的网络，然后接一个RE的网络，这种pipeline您有没有试过，准确率怎么样，有什么优缺点？能否回复一下，谢谢！

[layoutlmv3] SER and RE task combined into one model

1

Hi @HYPJUDY I have a question that do you try to combine SER and RE task into one model, that is to say, just use one model to support SER...