README.md

metadata

license: mit
base_model: alexdg19/bert_large_xsum_samsum
tags:
  - generated_from_trainer
datasets:
  - samsum
metrics:
  - rouge
model-index:
  - name: bert_large_xsum_samsum3
    results:
      - task:
          name: Sequence-to-sequence Language Modeling
          type: text2text-generation
        dataset:
          name: samsum
          type: samsum
          config: samsum
          split: test
          args: samsum
        metrics:
          - name: Rouge1
            type: rouge
            value: 0.5313

bert_large_xsum_samsum3

This model is a fine-tuned version of alexdg19/bert_large_xsum_samsum on the samsum dataset. It achieves the following results on the evaluation set:

• Loss: 2.2354
• Rouge1: 0.5313
• Rouge2: 0.2827
• Rougel: 0.4367
• Rougelsum: 0.4357
• Gen Len: 30.939

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
• gradient_accumulation_steps: 2
• total_train_batch_size: 4
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 10
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Rouge1	Rouge2	Rougel	Rougelsum	Gen Len
No log	1.0	164	1.1370	0.5599	0.3246	0.4748	0.4743	29.0122
No log	2.0	328	1.2659	0.5494	0.3033	0.4623	0.4612	27.0671
No log	3.0	492	1.4188	0.5198	0.2726	0.436	0.4346	28.6768
0.6603	4.0	656	1.5628	0.5391	0.2905	0.4555	0.4553	28.6159
0.6603	5.0	820	1.9045	0.5237	0.2774	0.4326	0.4321	31.5854
0.6603	6.0	984	2.0670	0.5199	0.2689	0.4251	0.4243	31.8049
0.1722	7.0	1148	1.9653	0.5269	0.2703	0.4342	0.4333	28.5122
0.1722	8.0	1312	2.1921	0.5296	0.2765	0.4393	0.4387	31.8354
0.1722	9.0	1476	2.4336	0.5299	0.2825	0.4399	0.4388	31.7988
0.052	10.0	1640	2.2354	0.5313	0.2827	0.4367	0.4357	30.939

Framework versions

• Transformers 4.35.0
• Pytorch 2.1.0+cu118
• Datasets 2.14.6
• Tokenizers 0.14.1