Link Search Menu Expand Document (external link)

Ranking and performance of all 572 ranked bert-base-uncased models (full table). The top 439 models were fully tested.

Notes:

  1. The baseline results can be found here
  2. While the average improvement is small, many datasets show large gains


  model_name avg mnli_lp 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
baseline bert-base-uncased 72.20 nan 83.05 89.59 65.92 46.95 68.96 64.38 81.83 49.45 78.16 89.70 68.53 91.58 69.07 83.73 81.99 59.97 66.68 89.88 90.27 84.85 59.98 91.97 52.80 85.86 96.06 68.33 36.01 79.91 52.85 67.76 85.37 69.48 63.25 50.56 62.12 72.32
1 ibm/ColD-Fusion-bert-base-uncased-itr23-seed0 75.64 75.73 85.12 89.13 66.26 49.09 74.37 76.79 81.98 58.00 78.20 90.73 84.10 92.10 69.30 84.38 87.01 59.36 85.58 89.27 90.56 89.40 77.98 94.15 55.52 88.57 97.20 81.00 36.28 81.07 55.05 68.37 85.00 70.30 65.83 52.11 62.50 71.30
2 ibm/ColD-Fusion-bert-base-uncased-itr24-seed0 75.55 76.42 85.06 89.10 65.98 48.50 74.43 76.79 81.50 62.00 78.57 90.44 81.60 92.02 69.69 83.84 86.52 60.17 84.62 90.02 90.55 89.77 78.34 93.46 57.19 89.12 96.60 81.40 35.94 81.63 53.67 67.73 85.00 69.45 66.14 47.89 63.46 71.60
3 ibm/ColD-Fusion-bert-base-uncased-itr22-seed0 75.45 77.00 85.26 88.80 66.26 47.50 74.22 78.57 81.40 59.00 78.53 90.65 84.00 92.07 69.75 84.41 86.03 60.77 82.69 89.40 90.37 89.68 77.98 93.69 55.88 88.93 97.20 81.00 35.88 81.98 51.28 69.26 85.35 69.42 65.83 49.30 62.50 71.23
4 ibm/ColD-Fusion-bert-base-uncased-itr14-seed0 75.38 76.96 84.77 89.27 66.16 48.34 74.19 76.79 81.78 59.00 78.30 90.60 83.90 92.06 69.75 84.27 88.48 60.11 85.58 89.75 90.77 88.46 75.45 94.04 55.25 89.12 97.20 80.00 35.90 81.00 54.44 67.73 84.88 69.54 66.61 45.07 63.46 71.57
5 ibm/ColD-Fusion-bert-base-uncased-itr26-seed0 75.22 74.47 84.94 89.23 66.10 48.69 74.25 80.36 82.17 52.00 78.43 90.81 83.50 92.25 68.58 84.38 86.76 59.53 83.65 90.13 90.84 90.15 78.70 94.84 56.43 88.95 97.60 81.60 35.97 80.30 55.45 67.98 84.65 69.31 65.05 39.44 63.46 71.40
6 ibm/ColD-Fusion-bert-base-uncased-itr28-seed0 75.21 77.79 84.84 89.23 65.92 49.28 74.25 78.57 81.98 57.00 78.53 90.66 83.30 92.12 69.04 83.99 87.01 59.57 84.62 90.26 90.76 90.15 77.62 94.61 57.29 89.01 97.00 81.60 36.06 81.42 52.69 67.60 84.88 68.98 65.20 38.03 63.46 71.13
7 ibm/ColD-Fusion-bert-base-uncased-itr21-seed0 75.21 76.97 84.64 89.07 66.04 49.34 73.46 76.79 81.88 58.00 78.50 90.58 82.60 91.96 69.56 83.82 85.29 58.17 85.58 89.64 90.44 89.12 77.62 94.27 55.93 88.86 97.00 80.80 36.17 82.13 54.07 68.62 85.35 69.37 65.52 43.66 61.54 72.03
8 ibm/ColD-Fusion-bert-base-uncased-itr27-seed0 75.19 77.24 84.69 89.23 65.70 48.47 74.19 73.21 81.21 57.00 78.50 90.41 83.20 92.05 68.84 83.84 87.75 59.26 84.62 89.82 90.74 90.24 78.70 94.27 56.47 89.00 96.80 81.80 36.30 79.80 54.65 68.24 85.70 69.37 66.30 42.25 63.46 70.77
9 ibm/ColD-Fusion-bert-base-uncased-itr18-seed0 75.18 73.37 85.13 89.07 66.08 48.41 73.91 75.00 82.07 58.00 78.07 90.36 81.80 91.97 68.84 83.92 86.27 58.40 85.58 89.02 90.60 89.77 71.48 94.04 56.65 88.79 97.40 80.60 36.43 81.07 53.74 68.37 85.12 69.34 64.89 50.70 63.46 72.13
10 ibm/ColD-Fusion-bert-base-uncased-itr20-seed0 75.13 75.99 85.02 89.03 66.04 48.56 74.01 78.57 81.78 55.00 78.53 90.33 83.40 91.99 69.56 83.95 85.78 58.17 83.65 89.93 90.40 89.87 78.70 93.12 55.48 88.68 97.40 80.40 36.35 81.63 54.38 68.11 85.12 69.55 66.93 42.25 61.54 71.37



Download full models ranking table: csv

Home

Table of contents