multinews_cnn_logs2

This model is a fine-tuned version of BeenaSamuel/t5_small_multi_news_abstractive_summarizer on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 1.7781
• Rouge1: 0.5231
• Rouge2: 0.1974
• Rougel: 0.4013
• Gen Len: 311.236

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: 0.0001
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 8
• total_train_batch_size: 64
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 500
• num_epochs: 5

Training results

Training Loss	Epoch	Step	Validation Loss	Rouge1	Rouge2	Rougel	Gen Len
1.9591	0.28	200	1.8011	0.519	0.1944	0.3973	311.236
1.8902	0.57	400	1.7998	0.5204	0.1946	0.3979	311.236
1.8851	0.85	600	1.7963	0.5203	0.1949	0.3981	311.236
1.9131	1.14	800	1.7947	0.52	0.1951	0.3985	311.236
1.929	1.42	1000	1.7919	0.5204	0.1955	0.3986	311.236
1.9045	1.71	1200	1.7881	0.5216	0.1957	0.3995	311.236
1.9542	1.99	1400	1.7881	0.5208	0.1959	0.3996	311.236
1.9129	2.28	1600	1.7842	0.5218	0.1965	0.4002	311.236
1.8727	2.56	1800	1.7848	0.5218	0.1965	0.4001	311.236
1.9194	2.85	2000	1.7833	0.5225	0.1968	0.4005	311.236
1.8275	3.13	2200	1.7821	0.5223	0.1968	0.4004	311.236
1.9338	3.42	2400	1.7809	0.5228	0.1971	0.4007	311.236
1.9234	3.7	2600	1.7809	0.5224	0.197	0.4008	311.236
1.904	3.98	2800	1.7795	0.5227	0.1972	0.4009	311.236
1.8844	4.27	3000	1.7791	0.5228	0.1973	0.4008	311.236
1.9315	4.55	3200	1.7788	0.5228	0.1972	0.4011	311.236
1.88	4.84	3400	1.7781	0.5231	0.1974	0.4013	311.236

Framework versions

• Transformers 4.39.3
• Pytorch 2.1.2
• Datasets 2.18.0
• Tokenizers 0.15.2