1tangerine1day/Sentiment-Analysis: Use Huggingface Bert ... - GitHub

Skip to content Dismiss alert {{ message }} 1tangerine1day / Sentiment-Analysis Public

• Notifications You must be signed in to change notification settings
• Fork 2
• Star 1

use huggingface bert/xlnet to do sentiment analysis on sst5 and amazon

1 star 2 forks Branches Tags Activity Star Notifications You must be signed in to change notification settings

• Code
• Issues
• Pull requests
• Actions
• Projects
• Security

  Uh oh!

  There was an error while loading. Please reload this page.

• Insights

Additional navigation options

1tangerine1day/Sentiment-Analysis  masterBranchesTagsGo to fileCodeOpen more actions menu

Folders and files

Name		Name	Last commit message	Last commit date
Latest commit   History 12 Commits
.ipynb_checkpoints		.ipynb_checkpoints
data		data
model		model
README.md		README.md
train.ipynb		train.ipynb
View all files

Repository files navigation

• README

Sentiment Analysis

Model

• Bert: https://hackmd.io/@h5v7rKYIT4GGOr9b0g0k7g/SyKjPsasv
• XLNet: https://hackmd.io/@h5v7rKYIT4GGOr9b0g0k7g/SyKjPsasv
• RoBerta: https://hackmd.io/@h5v7rKYIT4GGOr9b0g0k7g/H1osvVOnv

Dataset

• amazon clawer
• sst5 : https://github.com/prrao87/fine-grained-sentiment/tree/master/data/sst
• amazon 5 Electronics (to balance data): https://jmcauley.ucsd.edu/data/amazon/

Performance

0 -- f1	1 -- f1	2 -- f1
BERT	0.87	0.47	0.93
XLNet	0.89	0.51	0.94
RoBERTa(sum)	0.86	0.43	0.91
RoBERTa(pool)	0.89	0.48	0.93

case study

• A: Not as extreme as touted. (3/5)
• B: IT WAS DIFFERENT TYPE MOVIE BUT IT WAS GOOD. SUSPENSEFUL. (5/5)
• C: Not impressed with this movie, too many unknown variables. (1/5)
• D: The first half is outstanding. Lupita Nyongo is excellent as always -- shes a moody complex character which makes complete sense after the big reveal.Once that big reveal happens, suspending disbelief is difficult, even for a horror movie. It is almost as if they came up with the main premise, than spit-balled an explanation. The explanation / back story is so absurd it requires way too much exposition. Heres hoping Peele gets more grounded in his third act. (3/5)

A	B	C	D
BERT	0	2	0	2
XLNet	1	2	1	1
RoBERTa(sum)	1	1	0	2
RoBERTa(pool)	1	1	0	1

• class:
  • 0-> 1、2 (stars)
  • 1-> 3 (stars)
  • 2-> 4、5 (stars)

About

use huggingface bert/xlnet to do sentiment analysis on sst5 and amazon

Topics

jupyter-notebook pytorch bert huggingface xlnet

Resources

Readme

Uh oh!

There was an error while loading. Please reload this page.

Activity

Stars

1 star

Watchers

1 watching

Forks

2 forks Report repository

Releases

No releases published

Packages

No packages published

Languages

• Jupyter Notebook 99.4%
• Python 0.6%

You can’t perform that action at this time.