yubowen-ph / jointer Goto Github PK

Hi~, the paper claims that you use the datasets published by CopyRE. They use the datasets only annotating the last word, do you also follow this preprocessing setting? Or do you preprocess on the original dataset released by CopyRE and annotate the whole span? Thanks for your reply~

Hi! I've read your paper last year and it's really nice work. Good news that your paper got accepted to ECAI. Congratulations!

Also, thank you for releasing the code. But I noticed that the NYT-single dataset is missing in the repo. Would you provide the pre-processed data, or the pre-processing script? A reference link to external source is also helpful.

Thanks,
AL

作者有NYT-single的数据集吗

in the code, i can't see the softmax in the HE extraction to extracte subject head tag. there is a FC network which named nn.linear()?
can you tell me where is the softmax?

I find that split_by_length () function cannot be found in the file analyse. py. Do you have the specific content of this function.Thank you.

UnboundLocalError: local variable 'nearest_subj_start_position_for_each_token' referenced before assignment

Hi: could anyone tell me why set 4 and 2 in model.py line 188? Thankyou
self.subj_sublayer = submodel.SubjTypeModel(opt, 4opt['hidden_dim'], 2opt['hidden_dim'])

/pytorch/aten/src/THC/THCTensorScatterGather.cu💯 void THCudaTensor_gatherKernel(TensorInfo<Real, IndexType>, TensorInfo<Real, IndexType>, TensorInfo<long, IndexType>, int, IndexType) [with IndexType = unsigned int, Real = float, Dims = 3]: block: [4,0,0], thread: [501,0,0] Assertion indexValue >= 0 && indexValue < src.sizes[dim] failed.
/pytorch/aten/src/THC/THCTensorScatterGather.cu💯 void THCudaTensor_gatherKernel(TensorInfo<Real, IndexType>, TensorInfo<Real, IndexType>, TensorInfo<long, IndexType>, int, IndexType) [with IndexType = unsigned int, Real = float, Dims = 3]: block: [4,0,0], thread: [502,0,0] Assertion indexValue >= 0 && indexValue < src.sizes[dim] failed.
/pytorch/aten/src/THC/THCTensorScatterGather.cu💯 void THCudaTensor_gatherKernel(TensorInfo<Real, IndexType>, TensorInfo<Real, IndexType>, TensorInfo<long, IndexType>, int, IndexType) [with IndexType = unsigned int, Real = float, Dims = 3]: block: [4,0,0], thread: [503,0,0] Assertion indexValue >= 0 && indexValue < src.sizes[dim] failed.
/pytorch/aten/src/THC/THCTensorScatterGather.cu💯 void THCudaTensor_gatherKernel(TensorInfo<Real, IndexType>, TensorInfo<Real, IndexType>, TensorInfo<long, IndexType>, int, IndexType) [with IndexType = unsigned int, Real = float, Dims = 3]: block: [4,0,0], thread: [504,0,0] Assertion indexValue >= 0 && indexValue < src.sizes[dim] failed.
/pytorch/aten/src/THC/THCTensorScatterGather.cu💯 void THCudaTensor_gatherKernel(TensorInfo<Real, IndexType>, TensorInfo<Real, IndexType>, TensorInfo<long, IndexType>, int, IndexType) [with IndexType = unsigned int, Real = float, Dims = 3]: block: [4,0,0], thread: [505,0,0] Assertion indexValue >= 0 && indexValue < src.sizes[dim] failed.
/pytorch/aten/src/THC/THCTensorScatterGather.cu💯 void THCudaTensor_gatherKernel(TensorInfo<Real, IndexType>, TensorInfo<Real, IndexType>, TensorInfo<long, IndexType>, int, IndexType) [with IndexType = unsigned int, Real = float, Dims = 3]: block: [4,0,0], thread: [506,0,0] Assertion indexValue >= 0 && indexValue < src.sizes[dim] failed.
/pytorch/aten/src/THC/THCTensorScatterGather.cu💯 void THCudaTensor_gatherKernel(TensorInfo<Real, IndexType>, TensorInfo<Real, IndexType>, TensorInfo<long, IndexType>, int, IndexType) [with IndexType = unsigned int, Real = float, Dims = 3]: block: [4,0,0], thread: [507,0,0] Assertion indexValue >= 0 && indexValue < src.sizes[dim] failed.
/pytorch/aten/src/THC/THCTensorScatterGather.cu💯 void THCudaTensor_gatherKernel(TensorInfo<Real, IndexType>, TensorInfo<Real, IndexType>, TensorInfo<long, IndexType>, int, IndexType) [with IndexType = unsigned int, Real = float, Dims = 3]: block: [4,0,0], thread: [508,0,0] Assertion indexValue >= 0 && indexValue < src.sizes[dim] failed.
/pytorch/aten/src/THC/THCTensorScatterGather.cu💯 void THCudaTensor_gatherKernel(TensorInfo<Real, IndexType>, TensorInfo<Real, IndexType>, TensorInfo<long, IndexType>, int, IndexType) [with IndexType = unsigned int, Real = float, Dims = 3]: block: [4,0,0], thread: [509,0,0] Assertion indexValue >= 0 && indexValue < src.sizes[dim] failed.
/pytorch/aten/src/THC/THCTensorScatterGather.cu💯 void THCudaTensor_gatherKernel(TensorInfo<Real, IndexType>, TensorInfo<Real, IndexType>, TensorInfo<long, IndexType>, int, IndexType) [with IndexType = unsigned int, Real = float, Dims = 3]: block: [4,0,0], thread: [510,0,0] Assertion indexValue >= 0 && indexValue < src.sizes[dim] failed.
/pytorch/aten/src/THC/THCTensorScatterGather.cu💯 void THCudaTensor_gatherKernel(TensorInfo<Real, IndexType>, TensorInfo<Real, IndexType>, TensorInfo<long, IndexType>, int, IndexType) [with IndexType = unsigned int, Real = float, Dims = 3]: block: [4,0,0], thread: [511,0,0] Assertion indexValue >= 0 && indexValue < src.sizes[dim] failed.
Traceback (most recent call last):
File "train_chinese.py", line 124, in
loss = model.update(batch)
File "/home/ubuntu/jelly/JointER/models/model.py", line 70, in update
subj_start_logits, subj_end_logits, obj_start_logits, obj_end_logits = self.model(inputs, mask, nearest_subj_start_position_for_each_token, distance_to_nearest_subj_start, distance_to_subj, nearest_obj_start_position_for_each_token, distance_to_nearest_obj_start)
File "/usr/local/lib/python3.5/dist-packages/torch/nn/modules/module.py", line 541, in call
result = self.forward(*input, **kwargs)
File "/home/ubuntu/jelly/JointER/models/model.py", line 259, in forward
obj_start_logits, obj_end_logits = self.obj_sublayer(hidden, sentence_rep, subj_start_position, subj_end_position, mask, distance_to_subj, nearest_obj_start_position_for_each_token, distance_to_nearest_obj_start)
File "/usr/local/lib/python3.5/dist-packages/torch/nn/modules/module.py", line 541, in call
result = self.forward(*input, **kwargs)
File "/home/ubuntu/jelly/JointER/models/submodel.py", line 182, in forward
h0, c0 = self.zero_state(batch_size)
File "/home/ubuntu/jelly/JointER/models/submodel.py", line 160, in zero_state
return h0.cuda(), c0.cuda()
RuntimeError: CUDA error: device-side assert triggered

Hello, I'm studying your paper rencently. Your work is very great, but I didn't understand how the model handle the EPO situation from the paper's METHODOLOGY section. In fact, I think that tagging the relation label to indicate the start-end position in the TER stage can tag only one relation to a entity pair, so how to tag another relation for the same entity pair? Could you give more details about it, thanks a lot.

the code missed a file named schemas.json when i run the code, how i can get that file?

Hi @yubowen-ph ,

Thank you very much for releasing the data and your source code.
I already tried to reproduce the results as in Figure 3 in your paper. However, it is only the same as in Fig. 3 for two cases: Normal and EntityPairOverlap.
In case of SingleEntityOverlap, I checked the F1-score of OrderRL is only 69.4, and your model ETL-span is 74.7. However, in Fig. 03, you reported both of them with F1-score are larger than 80. Could you please give me the detailed F1-score in this figure? As I wonder about the case SEO in Figure 3.

Thank you very much for your help.

你好，请问代码中随机选择一个主体的代码的用意是什么？
比如一个句子有多个主体，但是只随机选择了一个，是不是意味着ETL只能从一个句子抽取一个关系，无法完成多关系抽取？

您好，我想问一下原生数据集怎么获取啊

@yubowen-ph Hi yubowen.
If it is convenient to share any of your trained model? Since I have no gpu to train yet, but i want to know the performance of the model myself.
Thanks a lot.

In your paper 2.3.1：‘character based word representations generated by running a CNN on the character sequence of xi.‘
Which paper can I refer to for implementation details ?

Hi! Sorry to bother you again. I am very interested in the work this paper. I want to replace the dataset with my own dataset, but I encountered the problem while processing my own data. Can you provide the code for preprocessing the dataset?

Thank you!