Comments (14)
Yes, but the impact on apc should be limited. This is an emprical conclusion and you can conduct experiments if you want.
from lcf-atepc.
Thank u for ur reply
I tested this model with a dataset in another language than English and Chinese. When I used the multilingual bert model I achieved high results, but when I used a monolingual model, I obtained very low results (F1-score = 0 for ATE task !!!), which is very weird. Normally the monolingual models are better than multilingual models as they have a larger number of vocabularies no?
Do u have any idea plz?
thank u
from lcf-atepc.
Which pretrained model dou use and can you share any visualization of this preoblem (e.g., code block)?
from lcf-atepc.
Note that this repo is hard coded to use BERTPretrainedModel and tokenizer, you may need to alter to use AutoModel and autotokenizer instead.
from lcf-atepc.
Hi,
I replaced the multilingual bert model by this model aubmindlab/bert-base-arabertv01 and I also used AutoModel and autotokenizer in ur code
As I said it gave me 0 for ATE and a low accuaracy for APC
Thank u
from lcf-atepc.
I dont have the dataset to debug, did you design the dataset as provided format? I received a similar report which is cuased by mis-annotation and label usage.
from lcf-atepc.
Yes, u were right; there was a problem with the data format. I fixed it, but the accuracy is still very low using the monolingual BERT model compared to the multilingual one.
I really cannot understand that because the monolingual models are generally better than multilingual ones
Do u have any idea plz?
thank u
from lcf-atepc.
Hi,
I suggest you share your code on Github so I can review it. otherwise I might have no idea where the problem comes from.
from lcf-atepc.
Thank u for ur effort to help us fixing errors. I am working on google colab. So I shared with u the notebook and the folder of code (my email address: [email protected]) to allow u to reproduce the results.
Thank u again for ur effort.
from lcf-atepc.
Do you solve the problem?
from lcf-atepc.
Hi,
Unfortunately, I am working on improving PyABSA, this repo is kind of out of maintance, you can try PyABSA which solve some problem about dataset. Or you can provide me with a cut of your dataset so I can analyze it.
from lcf-atepc.
I click the close button accidently, and look forward to your reply.
from lcf-atepc.
@Phd-Student2018 No not yet you?
from lcf-atepc.
There is no known error found in your data, maybe you can debug via pycharm, etc. To see what happened in tokenization (I suspect the problem is tokenization, or using incompatible tokenizer and model)
from lcf-atepc.
Related Issues (20)
- 关于使用额外的训练数据 HOT 3
- def get_ids_for_local_context_extractor HOT 1
- 关于数据转换。 HOT 12
- cross validation HOT 1
- Evaluation HOT 1
- The problem of the training on new dataset HOT 8
- pip install -U pyabsa问题 HOT 2
- 掩码长度问题 HOT 20
- 只有cpu可以训练吗? HOT 6
- 训练模型数据转换错误 HOT 3
- 你好,我试验了你们的这个多任务学习模型,有一些问题想请教。 HOT 3
- evaluation HOT 2
- Regard Dataset HOT 1
- Regrad loss function
- 预测的问题 HOT 1
- Hello, how can I solve this problem? thank you very much HOT 7
- 数据集标签问题 HOT 2
- 用我的数据集做预测时,遇到了一个问题 HOT 17
- LCF
- predicited lable
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
-
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
-
TensorFlow
An Open Source Machine Learning Framework for Everyone
-
Django
The Web framework for perfectionists with deadlines.
-
Laravel
A PHP framework for web artisans
-
D3
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
-
Recommend Topics
-
javascript
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
-
web
Some thing interesting about web. New door for the world.
-
server
A server is a program made to process requests and deliver data to clients.
-
Machine learning
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
-
Visualization
Some thing interesting about visualization, use data art
-
Game
Some thing interesting about game, make everyone happy.
Recommend Org
-
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
-
Microsoft
Open source projects and samples from Microsoft.
-
Google
Google ❤️ Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.
from lcf-atepc.