bart-email-multitask

This model is a fine-tuned version of rosypossyyyy/bart-email-multitask on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.5448
• Rouge1: 51.01
• Rouge2: 36.53
• Rougel: 44.46
• Rougelsum: 44.91
• Gen Len: 53.5864

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-06
• train_batch_size: 4
• eval_batch_size: 4
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 16
• optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 200
• num_epochs: 2
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Rouge1	Rouge2	Rougel	Rougelsum	Gen Len
0.3719	0.1726	500	0.5641	49.82	35.56	43.37	43.9	55.6862
0.39	0.3452	1000	0.5499	50.66	36.17	44.07	44.56	53.0616
0.4325	0.5178	1500	0.5412	50.59	36.52	44.19	44.63	54.4194
0.4328	0.6904	2000	0.5556	50.8	36.6	44.19	44.7	54.1698
0.4133	0.8630	2500	0.5358	50.67	36.36	44.02	44.56	54.8808
0.3576	1.0356	3000	0.5371	51.14	36.78	44.59	45.09	53.4543
0.3878	1.2082	3500	0.5359	50.77	36.52	44.17	44.73	54.4243
0.3757	1.3808	4000	0.5294	50.76	36.73	44.34	44.9	54.2312

Framework versions

• Transformers 4.57.5
• Pytorch 2.9.0+cu126
• Datasets 4.5.0
• Tokenizers 0.22.2