File size: 2,649 Bytes
7d4a760 12ff013 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 |
---
license: afl-3.0
---
## Model Recycling
[Evaluation on 36 datasets](https://ibm.github.io/model-recycling/model_gain_chart?avg=1.62&mnli_lp=nan&20_newsgroup=-3.07&ag_news=-0.36&amazon_reviews_multi=-0.59&anli=1.28&boolq=3.32&cb=11.52&cola=-0.74&copa=2.85&dbpedia=0.83&esnli=-0.12&financial_phrasebank=13.84&imdb=-0.18&isear=1.24&mnli=-0.15&mrpc=0.16&multirc=0.51&poem_sentiment=9.23&qnli=1.11&qqp=-0.07&rotten_tomatoes=-0.40&rte=4.87&sst2=-1.58&sst_5bins=-1.05&stsb=0.73&trec_coarse=-0.43&trec_fine=6.22&tweet_ev_emoji=-0.18&tweet_ev_emotion=0.40&tweet_ev_hate=-1.57&tweet_ev_irony=6.23&tweet_ev_offensive=0.40&tweet_ev_sentiment=-0.08&wic=-0.83&wnli=4.01&wsc=1.54&yahoo_answers=-0.63&model_name=Moussab%2Fdeepset_bert-base-cased-squad2-orkg-unchanged-5e-05&base_name=bert-base-cased) using Moussab/deepset_bert-base-cased-squad2-orkg-unchanged-5e-05 as a base model yields average score of 74.04 in comparison to 72.43 by bert-base-cased.
The model is ranked 3rd among all tested models for the bert-base-cased architecture as of 21/12/2022
Results:
| 20_newsgroup | ag_news | amazon_reviews_multi | anli | boolq | cb | cola | copa | dbpedia | esnli | financial_phrasebank | imdb | isear | mnli | mrpc | multirc | poem_sentiment | qnli | qqp | rotten_tomatoes | rte | sst2 | sst_5bins | stsb | trec_coarse | trec_fine | tweet_ev_emoji | tweet_ev_emotion | tweet_ev_hate | tweet_ev_irony | tweet_ev_offensive | tweet_ev_sentiment | wic | wnli | wsc | yahoo_answers |
|---------------:|----------:|-----------------------:|--------:|--------:|-----:|--------:|-------:|----------:|--------:|-----------------------:|-------:|--------:|--------:|--------:|----------:|-----------------:|--------:|--------:|------------------:|-------:|--------:|------------:|-------:|--------------:|------------:|-----------------:|-------------------:|----------------:|-----------------:|---------------------:|---------------------:|--------:|-------:|--------:|----------------:|
| 78.6644 | 88.7 | 65.12 | 47.8438 | 71.5902 | 75 | 81.1122 | 55 | 79.6 | 89.5155 | 82.2 | 90.968 | 69.6219 | 83.2384 | 83.0882 | 60.9736 | 76.9231 | 91.1038 | 89.8788 | 84.1463 | 67.509 | 89.9083 | 50.362 | 85.254 | 96.2 | 79.2 | 44.062 | 79.24 | 51.2121 | 71.4286 | 84.6512 | 68.1456 | 63.9498 | 56.338 | 63.4615 | 70.4 |
For more information, see: [Model Recycling](https://ibm.github.io/model-recycling/)
|