Evaluation results for ffgcc/InfoCSE-bert-base 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 1st among all tested models for the bert-base-uncased architecture as of 21/12/2022.


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

@@ -0,0 +1,17 @@
+# ffgcc/InfoCSE-bert-base model
+This model is based on bert-base-uncased pretrained model.
+
+
+## Model Recycling
+
+[Evaluation on 36 datasets](https://ibm.github.io/model-recycling/model_gain_chart?avg=2.08&mnli_lp=nan&20_newsgroup=-0.67&ag_news=-0.26&amazon_reviews_multi=0.42&anli=1.27&boolq=2.36&cb=7.05&cola=2.16&copa=11.55&dbpedia=-1.00&esnli=0.59&financial_phrasebank=15.07&imdb=-0.70&isear=2.70&mnli=0.60&mrpc=2.08&multirc=-1.37&poem_sentiment=8.32&qnli=1.26&qqp=0.40&rotten_tomatoes=0.98&rte=1.75&sst2=0.57&sst_5bins=1.46&stsb=1.12&trec_coarse=1.14&trec_fine=8.87&tweet_ev_emoji=0.81&tweet_ev_emotion=1.23&tweet_ev_hate=1.25&tweet_ev_irony=-2.33&tweet_ev_offensive=-0.02&tweet_ev_sentiment=1.02&wic=3.68&wnli=0.14&wsc=1.35&yahoo_answers=-0.12&model_name=ffgcc%2FInfoCSE-bert-base&base_name=bert-base-uncased) using ffgcc/InfoCSE-bert-base as a base model yields average score of 74.28 in comparison to 72.20 by bert-base-uncased.
+
+The model is ranked 1st among all tested models for the bert-base-uncased 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 |
+|---------------:|----------:|-----------------------:|--------:|--------:|--------:|--------:|-------:|----------:|--------:|-----------------------:|-------:|--------:|--------:|--------:|----------:|-----------------:|--------:|--------:|------------------:|--------:|--------:|------------:|--------:|--------------:|------------:|-----------------:|-------------------:|----------------:|-----------------:|---------------------:|---------------------:|--------:|--------:|--------:|----------------:|
+|        82.3818 |   89.3333 |                  66.34 | 48.2188 |  71.315 | 71.4286 | 83.9885 |     61 |   77.1667 | 90.2891 |                   83.6 | 90.872 | 71.7731 | 84.3267 | 84.0686 |   58.6015 |               75 | 91.1404 | 90.6752 |           85.8349 | 61.7329 | 92.5459 |     54.2534 | 86.9799 |          97.2 |        77.2 |            36.82 |              81.14 |         54.1077 |          65.4337 |              85.3488 |              70.4982 | 66.9279 | 50.7042 | 63.4615 |            72.2 |
+
+
+For more information, see: [Model Recycling](https://ibm.github.io/model-recycling/)