jialicheng/unlearn_samsum_t5-base_neggrad_6_42

samsum_42

This model is a fine-tuned version of google/t5-v1_1-base on the samsum dataset. It achieves the following results on the evaluation set:

• Loss: 1.7509
• Rouge1: 41.497
• Rouge2: 21.2823
• Rougel: 36.4325
• Rougelsum: 38.3568
• Gen Len: 12.3166
• Test Rougel: 36.4325
• Df Rougel: 35.9553
• Unlearn Overall Rougel: 0.7386
• Unlearn Time: 500.4764

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: 1e-05
• train_batch_size: 32
• eval_batch_size: 8
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 5

Training results

Training Loss	Epoch	Step	Validation Loss	Rouge1	Rouge2	Rougel	Rougelsum	Gen Len	Overall Rougel	Unlearn Overall Rougel	Time
No log	1.0	28	1.5108	48.53	24.364	41.2958	44.6482	21.4866	0.1667	0.1667	-1
No log	2.0	56	1.5695	45.176	23.127	39.3251	41.7401	15.7861	0.3573	0.3573	-1
No log	3.0	84	1.7509	41.497	21.2823	35.9553	38.3568	12.3166	0.7386	0.7386	-1
No log	4.0	112	1.9471	39.8092	20.0595	35.3111	36.7419	11.5990	0.3230	0.3230	-1
No log	5.0	140	2.0325	39.3494	19.7165	34.6824	36.3139	11.4156	0.4877	0.4877	-1

Framework versions

• Transformers 4.39.3
• Pytorch 2.3.0+cu121
• Datasets 2.19.0
• Tokenizers 0.15.2