isakzhang/generative-absa

generative-absa's Issues

算力问题

您好，请问你们当时跑实验的时候采用的算力是多少呢，之后我想知道跑这个实验要求的最低算力是多少，我现在在用的是32G的v100显卡，总是卡在epoch0那里，请问这里是不是需要我提升算力呢

code is incomplete in main.py line 332。

the evaluate func is define as : def evaluate(data_loader, model, paradigm, task, sents):
but parameter not passed in :

Generative-ABSA/main.py

Line 332 in 8f4671b

dev_result = evaluate(dev_loader, model, args.paradigm, args.task)

It seems that the eval process is not finished, can you add the complete code?

14res结果无法复现

您好，在对您论文得到的结果进行复现时，对于extraction-style上14res的结果（我设置了多个随机种子进行运行），均与您论文结果相差较大

can i use this model on other language?

i want to train a Chinese language model,how can i train on Chinese?

Problems

Hello Isak. I have encountered the following problems during the debugging process, how can i solve them? Thanks a lot.
Problem: RuntimeError: stack expects each tensor to be equal size, but got [33] at entry 0 and [22] at entry 1

hparams error

hello Do you know this error?
Traceback (most recent call last):
File "D:\sentiment\GAS\Generative-ABSA-main\main.py", line 264, in
model = T5FineTuner(args)
^^^^^^^^^^^^^^^^^
File "D:\sentiment\GAS\Generative-ABSA-main\main.py", line 89, in init
self.hparams = hparams
^^^^^^^^^^^^
File "D:\anaconda3\Lib\site-packages\torch\nn\modules\module.py", line 1674, in setattr
super().setattr(name, value)
AttributeError: property 'hparams' of 'T5FineTuner' object has no setter

about Prediction Normalization

hello， Where can I find the code for the "Prediction Normalization" part, I don't seem to find it

GPU or CPU

Thanks for your code! I have a question, do I have to use a GPU for training? Can I use a CPU? Hope to get your reply soon.