xingying-stack / milr Goto Github PK

soft_rule_regularization.py中load_and_prune函数：
with open("*.pl", "rb") as file:
rule_miner = pickle.load(file)
一直会因为rule_dwie.txt的数据格式报错，UnpicklingError: invalid load key, '['.

Hello, I encountered an "index out of sequence length" issue while trying to replicate your code using the Docred dataset. Have you encountered such an issue before, and any help would be greatly appreciated.

请问你使用过别的数据集吗，除了docred和dwie之外的数据集，如semeval。我处理了semeval的数据集格式，但是总是在fact后，train的时候，报出错误。
**
Traceback (most recent call last):
File "MILR-main/train.py", line 381, in
main()
File "MILR-main\model.py", line 35, in init
self.srr = SRR(rule_path=rule_path, prior_path=prior_path, minC=minC, temperature=temperature,
File "MILR-main\soft_rule_regularization.py", line 59, in init
self.eliminate_anti_rule_for_ILP()
File "MILR-main\soft_rule_regularization.py", line 247, in eliminate_anti_rule_for_ILP
self.confidence_1_hop = torch.stack(confi_1_hop, dim=0) # stack expects a non-empty TensorList
RuntimeError: stack expects a non-empty TensorList
**
也不知道是什么原因导致的，在fact构建的时候还是正常的