Comments (2)
By default we treat each line as a separate example, meaning each line in the targets needs to align with each line in the predictions. Rouge scores are computed on each pair of examples before being aggregated. Removing the check will mean any extra lines will be discarded, but I imagine the alignment will still be off, producing the low scores you're seeing.
It sounds like for your use case you might want to compute rouge between entire files, rather than treating each line separately? In this case you could try adding the flag --delimiter=++EOF++ (using any string not appearing in the files as a delimiter), so it will treat each file as a single example.
If your use case is different let me know. Some sample inputs and expected outputs might help in this case.
from google-research.
May I ask how do you mean by "each line in the targets needs to align with each line in the predictions". Does it mean that for every system summary there should be only one gold summary. The rouge measure can essentially evaluate a system summary against multiple gold summaries. Does it mean that your package only support single gold summary?!
I am wondering if it is possible to provide more detail information on using your rouge package ? Should I call your package similar to pyrouge (https://pypi.org/project/pyrouge/)? ( I have copied the highlights below)
"To automatically match a system summary with the corresponding model summaries, pyrouge uses regular expressions. For example, let’s assume your system summaries are named with a combination of a fixed name and a variable numeric ID like this:
some_name.001.txt
some_name.002.txt
some_name.003.txt
…
and the model summaries like this, with uppercase letters identifying multiple model summaries for a given document:
some_name.A.001.txt
some_name.B.001.txt
some_name.C.001.txt
some_name.A.002.txt
some_name.B.002.txt
…
The group in the system_filename_pattern tells pyrouge which part of the filename is the ID – in this case (\d+). You have to use round brackets to indicate a group, or else pyrouge won’t be able to tell apart the ID from the rest of the filename. pyrouge then uses that ID to find all matching model summaries. The special placeholder #ID# tells pyrouge where it should expect the ID in the model_filename_pattern. The [A-Z] part matches multiple model summaries for that ID."
from google-research.
Related Issues (20)
- About the Synthetic Repetition HOT 3
- i3d model's input range for FVD calculation
- Finetuning deplot
- Redundancy of jnp.where operation in d3pm p_logits.
- IABOT
- [email protected]
- Proposal for Collaborative Research on New Supervised Contrastive Learning Loss Functions
- TFT is not sorting based on dates automatically.
- I want to know the AI basics
- your code
- Your code made me vomit
- Meth Whores down to fuckb in the greater flint michigan
- [HITNET] How can I train my own models?
- AttributeError: module 'tensorflow.compat.v1' has no attribute 'BinaryCrossentropy'
- rep
- flint town
- which version of keras is used by latest kws_streaming HOT 3
- readme message error
- Save me HOT 2
- issue with the protnlm_use_model_for_inference_uniprot_2022_04.ipynb colab notebook
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 google-research.