Evaluation results for mariolinml/deberta-v3-base_MNLI_10_19_v0 model as a base model for other tasks

As part of a research effort to identify high quality models in Huggingface that can serve as base models for further finetuning, we evaluated this by finetuning on 36 datasets. The model ranks 3rd among all tested models for the microsoft/deberta-v3-base architecture as of 23/01/2023.


To share this information with others in your model card, please add the following evaluation results to your README.md page.

For more information please see https://ibm.github.io/model-recycling/ or contact me.

Best regards,
Elad Venezian
[email protected]
IBM Research AI

README.md

@@ -50,3 +50,17 @@ The following hyperparameters were used during training:
 - Pytorch 1.12.1+cu113
 - Datasets 2.6.1
 - Tokenizers 0.13.1
+
+## Model Recycling
+
+[Evaluation on 36 datasets](https://ibm.github.io/model-recycling/model_gain_chart?avg=0.71&mnli_lp=nan&20_newsgroup=-0.57&ag_news=-0.21&amazon_reviews_multi=-0.12&anli=1.28&boolq=-1.15&cb=7.14&cola=-1.72&copa=10.60&dbpedia=0.00&esnli=-0.81&financial_phrasebank=2.42&imdb=-0.12&isear=-0.48&mnli=-0.06&mrpc=-0.97&multirc=2.12&poem_sentiment=1.73&qnli=0.25&qqp=0.08&rotten_tomatoes=-0.65&rte=3.21&sst2=0.12&sst_5bins=0.48&stsb=1.46&trec_coarse=-0.16&trec_fine=0.78&tweet_ev_emoji=-0.67&tweet_ev_emotion=0.29&tweet_ev_hate=-0.22&tweet_ev_irony=0.03&tweet_ev_offensive=-0.76&tweet_ev_sentiment=-0.54&wic=-1.15&wnli=4.44&wsc=-0.63&yahoo_answers=0.10&model_name=mariolinml%2Fdeberta-v3-base_MNLI_10_19_v0&base_name=microsoft%2Fdeberta-v3-base) using mariolinml/deberta-v3-base_MNLI_10_19_v0 as a base model yields average score of 79.75 in comparison to 79.04 by microsoft/deberta-v3-base.
+
+The model is ranked 3rd among all tested models for the microsoft/deberta-v3-base architecture as of 22/01/2023
+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 |
+|---------------:|----------:|-----------------------:|--------:|--------:|--------:|--------:|-------:|----------:|--------:|-----------------------:|-------:|--------:|--------:|--------:|----------:|-----------------:|-------:|--------:|------------------:|--------:|--------:|------------:|--------:|--------------:|------------:|-----------------:|-------------------:|----------------:|-----------------:|---------------------:|---------------------:|--------:|--------:|--------:|----------------:|
+|        85.8471 |   90.2333 |                  66.74 | 60.0625 | 81.8349 | 82.1429 | 84.8514 |     69 |   79.4333 | 91.1136 |                   86.9 | 94.372 |  71.382 | 89.7172 | 88.2353 |   64.3771 |          88.4615 | 93.758 | 91.8699 |           89.7749 | 85.5596 | 95.1835 |     57.4661 | 91.7396 |          97.6 |        91.8 |           45.526 |            84.2365 |         55.9933 |          79.8469 |              84.3023 |              71.2634 | 70.0627 | 74.6479 | 63.4615 |         72.1333 |
+
+
+For more information, see: [Model Recycling](https://ibm.github.io/model-recycling/)