Edit model card

bart-cnn-pubmed-arxiv-pubmed-arxiv-earlystopping

This model is a fine-tuned version of theojolliffe/bart-cnn-pubmed-arxiv-pubmed-arxiv on an unknown dataset. It achieves the following results on the evaluation set:

  • Loss: 0.8793
  • Rouge1: 56.2055
  • Rouge2: 41.9231
  • Rougel: 45.0616
  • Rougelsum: 54.6643
  • Gen Len: 142.0

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

  • learning_rate: 2e-05
  • train_batch_size: 2
  • eval_batch_size: 2
  • seed: 42
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • num_epochs: 1000
  • mixed_precision_training: Native AMP

Training results

Training Loss Epoch Step Validation Loss Rouge1 Rouge2 Rougel Rougelsum Gen Len
No log 0.31 125 1.2057 50.9339 30.6777 32.6396 47.9592 141.3519
No log 0.63 250 1.0933 52.0728 31.2361 32.8214 48.9776 141.9815
No log 0.94 375 0.9685 51.6847 32.1578 34.1933 48.8808 141.5556
1.1594 1.26 500 0.9725 50.5131 30.6043 32.1861 47.4346 142.0
1.1594 1.57 625 0.9342 52.228 32.2073 33.797 49.2395 142.0
1.1594 1.88 750 0.8715 52.2 33.6602 36.1303 49.7138 141.6481
1.1594 2.2 875 0.8334 53.116 33.9871 35.9641 50.7658 141.8889
0.6845 2.51 1000 0.8241 52.2612 32.8025 35.27 49.5694 142.0
0.6845 2.83 1125 0.7986 54.1803 35.0019 37.4582 51.4577 142.0
0.6845 3.14 1250 0.8532 52.1328 32.6086 34.7455 49.6219 141.7037
0.6845 3.45 1375 0.8319 51.9614 32.8544 35.3269 49.3279 141.7593
0.4488 3.77 1500 0.8033 53.1404 34.6086 37.5482 50.7414 142.0
0.4488 4.08 1625 0.8322 53.1736 34.8662 37.7514 51.0601 142.0
0.4488 4.4 1750 0.7985 51.8251 32.9457 36.4164 49.55 142.0
0.4488 4.71 1875 0.8049 54.3423 36.6293 39.1316 52.2706 141.8148
0.3017 5.03 2000 0.8148 53.0698 35.2569 38.406 50.9346 141.7778
0.3017 5.34 2125 0.8153 53.4479 35.1525 37.8071 51.3731 141.0741
0.3017 5.65 2250 0.8009 52.5517 34.8287 37.999 50.2889 141.6111
0.3017 5.97 2375 0.7509 54.2725 37.4164 40.516 52.1379 142.0
0.2052 6.28 2500 0.8019 54.622 36.4776 39.9306 52.5069 142.0
0.2052 6.6 2625 0.8176 55.4796 38.4502 41.5523 53.5211 142.0
0.2052 6.91 2750 0.7956 55.4906 37.9064 40.845 53.107 141.9815
0.2052 7.22 2875 0.7966 54.5177 37.3399 40.7678 52.4241 142.0
0.1465 7.54 3000 0.8311 54.3473 37.0659 40.2507 52.372 142.0
0.1465 7.85 3125 0.8227 53.9245 36.4695 39.1205 51.9416 141.8889
0.1465 8.17 3250 0.7947 54.766 38.4275 41.2293 52.9075 142.0
0.1465 8.48 3375 0.7954 54.5305 37.6934 40.6804 52.5884 141.9444
0.115 8.79 3500 0.8433 54.7962 37.9373 41.3906 52.3778 142.0
0.115 9.11 3625 0.8416 56.59 41.2271 44.4207 54.7199 142.0
0.115 9.42 3750 0.8164 55.1903 39.0588 41.4908 53.4897 142.0
0.115 9.74 3875 0.8363 55.2894 39.3598 42.1138 53.831 141.8889
0.0912 10.05 4000 0.8850 55.7705 40.4924 43.1048 54.254 142.0
0.0912 10.36 4125 0.8268 56.1664 40.641 42.798 54.0001 141.9259
0.0912 10.68 4250 0.8564 55.4701 39.4949 42.2559 53.4486 141.8889
0.0912 10.99 4375 0.8557 56.0849 41.2861 45.8277 54.5999 141.6667
0.0707 11.31 4500 0.8432 54.9496 39.3006 42.0025 53.3854 142.0
0.0707 11.62 4625 0.8377 54.2438 37.6959 40.4637 52.3088 142.0
0.0707 11.93 4750 0.8794 55.9488 40.5401 43.7347 54.1282 142.0
0.0707 12.25 4875 0.8563 57.8762 43.366 46.6757 56.6985 142.0
0.0604 12.56 5000 0.8835 54.8926 39.3755 42.384 53.2687 141.6481
0.0604 12.88 5125 0.8570 55.6656 39.849 42.1455 54.352 142.0
0.0604 13.19 5250 0.8539 57.1549 41.901 45.153 55.213 142.0
0.0604 13.51 5375 0.8847 56.3279 40.9269 43.416 54.7242 142.0
0.051 13.82 5500 0.8795 56.8982 42.3333 45.2669 55.1034 142.0
0.051 14.13 5625 0.8751 55.3173 40.2853 43.2479 53.7236 142.0
0.051 14.45 5750 0.8799 56.1678 41.0862 43.8581 54.6316 142.0
0.051 14.76 5875 0.8678 57.3539 43.0473 44.8511 55.6474 142.0
0.0467 15.08 6000 0.8945 56.1939 41.985 45.0266 54.8139 142.0
0.0467 15.39 6125 0.9245 56.2071 41.5265 44.3228 54.5042 141.4074
0.0467 15.7 6250 0.8793 56.2055 41.9231 45.0616 54.6643 142.0

Framework versions

  • Transformers 4.19.1
  • Pytorch 1.11.0+cu113
  • Datasets 2.2.1
  • Tokenizers 0.12.1
Downloads last month
3