learn3r/longt5_xl_summ_screen_memsum_25

longt5_xl_summ_screen_25

This model is a fine-tuned version of longt5_xl_summ_screen_memsum_20/checkpoint-140 on the learn3r/summ_screen_memsum_oracle dataset. It achieves the following results on the evaluation set:

• Loss: 4.0742
• Rouge1: 39.5624
• Rouge2: 10.2833
• Rougel: 21.2004
• Rougelsum: 34.0767
• Gen Len: 152.0325

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: 0.001
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 32
• total_train_batch_size: 256
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: constant
• num_epochs: 5.0

Training results

Training Loss	Epoch	Step	Validation Loss	Rouge1	Rouge2	Rougel	Rougelsum	Gen Len
0.1846	0.97	14	4.6151	37.719	9.6532	20.2955	32.3806	100.9527
0.1441	1.95	28	4.1640	36.8632	9.6545	20.9349	31.8954	105.5799
0.1379	2.99	43	4.0742	39.5624	10.2833	21.2004	34.0767	152.0325
0.089	3.97	57	4.5216	40.2528	10.9254	21.6978	34.6793	176.0976
0.1028	4.87	70	4.1434	32.7739	9.1305	19.6246	27.9012	59.6775

Framework versions

• Transformers 4.34.0.dev0
• Pytorch 2.0.1+cu117
• Datasets 2.14.5
• Tokenizers 0.13.3