isakzhang/absa-quad

absa-quad's Issues

Which python version?

Thanks for sharing library versions. Could you please let us know the python version as well? The code breaks with python 3.10

代码问题

您好，非常感谢您提供的源代码。我是这个领域的小白，正在接触。复现过程中，代码显示好多错误，是我没设置好的问题吗？例如：TypeError: read_line_examples_from_file() missing 1 required positional argument: 'silence'。请教一下

您好，非常感谢您提供的源代码。

您好，非常感谢您提供的源代码。我是小白，在复现过程中代码显示有好多错误，是我没有设置好参数的问题吗？例如：TypeError: read_line_examples_from_file() missing 1 required positional argument: 'silence'

The purpose of code "lm_labels[lm_labels[:, :] == self.tokenizer.pad_token_id] = -100 "

Hi，

I'm trying to reimplement your work, and i'm curious about the following code:
lm_labels[lm_labels[:, :] == self.tokenizer.pad_token_id] = -100 (main.py, Line 109)

what's the purpose of this?

Model checkpoint

您好，非常感谢您出色的工作成果。
想请问下，可以提供模型训练好的checkpoint吗，我在使用您的代码时发现性能和论文中显示的效果差距非常大。

An error occurs when loading a previously trained model from 'do_direct_eval()' for evaluation.

Hello, thanks for sharing your code.

Currently, model training has been performed through the 'do_train()' code, and the trained model has been saved. After that, I uncommented the commented part to load the trained model for inference, but an error occurred. The content of the error is as follows.

****** Conduct Evaluating with the last state ******
Traceback (most recent call last):
File "myPath\ABSA-QUAD\main.py", line 312, in
model = T5FineTuner(args)
TypeError: init() missing 2 required positional arguments: 'tfm_model' and 'tokenizer'

Please let me know how to solve this.

Address issues that arise during the training of the repository.

Hello, thanks for sharing your code.

Currently, when I run this repository, I encounter some issues, and I have provided solutions that seem to work well. I am sharing this with everyone in the hope of receiving approval from the author.

1. silent in read_line_examples_from_file isn't set default:
   Solve:

def read_line_examples_from_file(data_path, silence=False):
    .....

2. Can't assign hparams in model T5FineTunerwith new version lightning >= 2.0.0:
   Solve:

self.hparams.update(vars(hparams))

3. New version lightning isn't support training_epoch_end and training_epoch_end, so adding prefix on_ before name of these functions

4. Lightning has integrated the gradient step in the 20 hooks, so I commented out the optimizer_step function because I observed issues with the optimizer closure."

5. Due to the internal separation of train and validation data within the model, it is not possible to assign output parameters to on_validation_epoch_end. Therefore, I removed it and replaced it with the following code snippet:

 class MyLightningModule(L.LightningModule):
    def __init__(self):
        super().__init__()
        self.validation_step_outputs = []

     def validation_step(self, ...):
         loss = ...
         self.validation_step_outputs.append(loss)
         return loss

    def on_validation_epoch_end(self):
        epoch_average = torch.stack(self.validation_step_outputs).mean()
        self.log("validation_epoch_average", epoch_average)
        self.validation_step_outputs.clear()  # free memory

6. Param gpus does not exist in the Lightning Trainer; instead, add the device parameter with the value 'auto' to automatically detect the available GPUs, and set accelerator='gpu'."
   Solve:

    train_params = dict(
        default_root_dir=args.output_dir,
        accumulate_grad_batches=args.gradient_accumulation_steps,
        devices='auto',
        gradient_clip_val=1.0,
        max_epochs=args.num_train_epochs,
        callbacks=[LoggingCallback()],
        accelerator='gpu'
    )

These are the solutions I have referenced online. If there are any errors, please overlook them and feel free to provide additional feedback

About the dataset

你好，我想把你的数据集应用在其他模型中，比如使用BIO标注方式进行方面词和观点词的抽取，这需要数据集中明确方面词和观点词的index。请问你们在标注数据集的时候有保留这部分数据嘛？