jialicheng/unlearn_samsum_t5-base_salun_2_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: 1.4604
• Rouge1: 49.7278
• Rouge2: 25.218
• Rougel: 41.5011
• Rougelsum: 45.8754
• Gen Len: 24.1042
• Test Rougel: 41.4686
• Df Rougel: 42.5366
• Unlearn Overall Rougel: -0.0340
• Unlearn Time: 3176.2771

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: 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	460	1.4833	49.3407	24.6019	42.6682	45.3389	23.3382	-0.3831	-0.3831	-1
No log	2.0	920	1.4752	49.3485	24.6668	43.2596	45.5003	23.9216	-0.6428	-0.6428	-1
1.8304	3.0	1380	1.4623	49.4331	25.0928	42.7932	45.4628	23.7279	-0.3018	-0.3018	-1
1.8304	4.0	1840	1.4638	49.6255	25.0168	42.6013	45.7687	24.3174	-0.1592	-0.1592	-1
1.7441	5.0	2300	1.4604	49.7278	25.218	42.5366	45.8754	24.1042	-0.0340	-0.0340	-1

Framework versions

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