<!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. -->
bart-large-cnn-small-xsum-5epochs
This model is a fine-tuned version of facebook/bart-large-cnn on the xsum dataset. It achieves the following results on the evaluation set:
- Loss: 2.7051
- Rouge1: 0.2859
- Rouge2: 0.0937
- Rougel: 0.2033
- Rougelsum: 0.2101
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: 2.045e-05
- train_batch_size: 4
- eval_batch_size: 8
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 16
- num_epochs: 5
- mixed_precision_training: Native AMP
Training results
Training Loss | Epoch | Step | Validation Loss | Rouge1 | Rouge2 | Rougel | Rougelsum |
---|---|---|---|---|---|---|---|
2.5007 | 0.32 | 16 | 2.0311 | 0.2393 | 0.0609 | 0.1618 | 0.1832 |
2.0942 | 0.64 | 32 | 1.9169 | 0.2906 | 0.1053 | 0.2072 | 0.2166 |
1.7543 | 0.96 | 48 | 1.9069 | 0.2904 | 0.0955 | 0.2058 | 0.2187 |
1.2476 | 1.28 | 64 | 1.9614 | 0.2928 | 0.1043 | 0.2081 | 0.2257 |
1.2318 | 1.6 | 80 | 1.9622 | 0.2892 | 0.0976 | 0.2099 | 0.2245 |
1.0768 | 1.92 | 96 | 2.0244 | 0.2935 | 0.1008 | 0.2095 | 0.2209 |
0.8845 | 2.24 | 112 | 2.0605 | 0.2886 | 0.0992 | 0.2039 | 0.2146 |
0.5722 | 2.56 | 128 | 2.2340 | 0.2852 | 0.0946 | 0.1983 | 0.2146 |
0.7132 | 2.88 | 144 | 2.1948 | 0.2838 | 0.0961 | 0.2047 | 0.2163 |
0.4438 | 3.2 | 160 | 2.3758 | 0.2869 | 0.0906 | 0.1987 | 0.2102 |
0.4194 | 3.52 | 176 | 2.5609 | 0.2882 | 0.0916 | 0.2022 | 0.2133 |
0.3404 | 3.84 | 192 | 2.4988 | 0.2884 | 0.0907 | 0.2022 | 0.213 |
0.2929 | 4.16 | 208 | 2.5802 | 0.2885 | 0.0967 | 0.2046 | 0.2141 |
0.2466 | 4.48 | 224 | 2.6590 | 0.2823 | 0.094 | 0.1994 | 0.2119 |
0.1889 | 4.8 | 240 | 2.7051 | 0.2859 | 0.0937 | 0.2033 | 0.2101 |
Framework versions
- Transformers 4.25.1
- Pytorch 1.13.0+cu117
- Datasets 2.7.1
- Tokenizers 0.13.2