scofield7419 / hesyfu Goto Github PK

Is it possible to share a sample of the dataset that was used in read_data?

Also, how may i ablate the constituency and dependency parts from the HeSyFu layers?

Lastly, why is it that the HeSyFu laters have dimension up to 350? but the output hidden representation(after coming out of the transformer_encoder layer has a dimension of 1068 in the last dimension)?

Hi, as the title, where is the requirements.txt file. So, I can know the environment you used, like the PyTorch version and so on.

Kind creator, I cannot understand the format of the datasets you use, i.e. input and output of the read_data, in this line of code in run_srl.py :

train, train_data_file, w_c_to_idx, c_c_to_idx, dep_lb_to_idx = read_data(train_file, {}, {}, {})

Looking on the conll2005/09/12 sites indicated, I cannot find an example compatible with the function.

Can you please provide a small example (one or two sentences) of "train_file"
or
an example of the output of the read_data function?