Comments (2)
The fact the loss doesn't change at all seems concerning. The overall approach should work as well for mbart, so there has to be some issue with training. I suggest you debug the loss, see if it returns any NaN. This may be caused by some mishandling of special tokens. Learning rate may also need to be changed. Also remember that mbart expects the language token at the start of the input/output.
But again, if loss is not changing, I would focus on that rather than checking the predictions.
from rebel.
It seems this has become stale, hence I am closing the issue. Feel free to share what the issue was if you found it or reopen it in case you have more details.
from rebel.
Related Issues (20)
- Documentation request for complete spacy setup HOT 1
- SpaCy component mapping to docs HOT 1
- error in your demo code HOT 2
- Issue while loading the trained checkpoint HOT 3
- How do I know which original span a predicted entity refers to? HOT 2
- DocRED dataset HOT 1
- Replicating REBEL from BART and some issues HOT 2
- Role of shift_tokens_left HOT 1
- Guide to Fine-tuning on Spacy HOT 2
- is it possible to specify which word i want to generate relations for? HOT 1
- Can't find factory for 'rebel' for language English (en). HOT 1
- Explainability of REBEL HOT 1
- Fine tuning for person to person entity relationship extraction HOT 1
- Problem with negative samples HOT 3
- Extraction of non-existant relation HOT 1
- Issue while running the default_model on training with conl dataset HOT 2
- Error while executing conl dataset HOT 1
- Dataset generation error HOT 1
- Pred file and gold file issues
- version incompatible HOT 2
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 rebel.