Comments (3)
Hi Peiyi,
The y-axis is in log-scale so it's still a long tail distribution. But you are right in the fact that in the chunking process, some distant arguments may not appear. This is a model limitation and I suppose long-context LMs (such as Longformers) could improve in this case.
from gen-arg.
I wonder if since the model is using BART, we may simply switch up to 1024 max length without much re-engineering?
And if we want to use this model for even longer documents, would it be possible to simply switch out the encoder for allenai's Longformer EncoderDecoder model?
from gen-arg.
I wonder if since the model is using BART, we may simply switch up to 1024 max length without much re-engineering?
And if we want to use this model for even longer documents, would it be possible to simply switch out the encoder for allenai's Longformer EncoderDecoder model?
Yes, you can change the base model BART to another encoder-decoder model that handles long context. The hard-coded max length is only to speed up training on the datasets that I used for the paper.
from gen-arg.
Related Issues (20)
- Access denied when downloading checkpoints from S3 HOT 2
- Testing the checkpoint on WikiEvents dataset HOT 2
- Only 10 F1 score on wikievent dataset HOT 13
- The event arguments annotation. HOT 1
- The head word F1 HOT 1
- TapKey Model Missing HOT 3
- Multiple arguments of the same argument role HOT 1
- data process HOT 6
- Share pretrained class vectors and tagger checkpoints
- Clarification needed on the implementation of Equation 4 of the paper HOT 1
- the implementation part of clarification in code? HOT 3
- 关于生成时约束词表 HOT 3
- convert_pointer_logits_to_lm_logits函数中的fill_value=-1000是怎么选取的? HOT 1
- About the evaluation set of ACE HOT 1
- 你好想问一下关于任务和数据的问题。 HOT 2
- 139 event types in paper vs 149 in csv file HOT 2
- download the checkpoint without aws account HOT 1
- lose RAMS dataset file? HOT 2
- how to generate event_role_ACE.json and event_role_KAIROS.json HOT 1
- Downloading checkpoint from s3
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 gen-arg.