distilbart-cnn-12-6-finetuned-resume-summarizer

This model is a fine-tuned version of Ameer05/model-tokenizer-repo on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 2.1123
• Rouge1: 52.5826
• Rouge2: 34.3861
• Rougel: 41.8525
• Rougelsum: 51.0015

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: 5e-05
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 32
• 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
No log	0.91	5	3.2243	42.8593	24.8652	34.1789	41.406
No log	1.91	10	2.6948	48.8571	28.6711	39.2648	46.188
No log	2.91	15	2.4665	50.6085	30.4034	39.7406	48.5449
No log	3.91	20	2.3329	52.2357	32.3398	41.574	49.4316
3.6611	4.91	25	2.2362	52.0134	33.1612	41.3103	50.255
3.6611	5.91	30	2.1833	51.5434	32.7045	40.5683	49.4238
3.6611	6.91	35	2.1462	53.5144	35.4518	42.8615	51.4053
3.6611	7.91	40	2.1518	52.0985	33.6754	41.5936	50.5159
2.0326	8.91	45	2.1075	53.1401	34.9721	42.2973	51.8454
2.0326	9.91	50	2.1123	52.5826	34.3861	41.8525	51.0015

Framework versions

• Transformers 4.15.0
• Pytorch 1.9.1
• Datasets 2.0.0
• Tokenizers 0.10.3