longt5_xl_sfd_bp_15

This model is a fine-tuned version of google/long-t5-tglobal-xl on the learn3r/summ_screen_fd_bp dataset. It achieves the following results on the evaluation set:

• Loss: 2.5840
• Rouge1: 29.7482
• Rouge2: 12.0072
• Rougel: 21.348
• Rougelsum: 28.5849
• Gen Len: 503.5769

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: 15.0

Training results

Training Loss	Epoch	Step	Validation Loss	Rouge1	Rouge2	Rougel	Rougelsum	Gen Len
2.5763	0.97	14	2.5415	10.6052	1.4494	10.4593	10.4801	509.6479
1.8998	1.95	28	1.7398	16.7989	4.1457	16.4049	15.1803	511.0
1.6403	2.99	43	1.5457	18.4716	5.4633	17.1393	16.9242	511.0
1.5012	3.97	57	1.5736	18.2259	5.3524	17.0162	16.7948	511.0
1.248	4.94	71	1.5482	20.8275	6.7412	18.0859	19.3113	511.0
1.0176	5.98	86	1.6254	21.1937	6.8813	18.411	19.8577	510.6775
0.8472	6.96	100	1.6212	26.1873	9.1581	20.393	24.1393	479.9704
0.7242	8.0	115	1.7231	23.5881	7.8961	18.7014	22.2999	506.9112
0.5876	8.97	129	1.9401	32.1851	12.6426	22.8358	30.6718	451.6982
0.4756	9.95	143	1.9001	31.353	12.994	23.1542	29.8375	455.5947
0.4042	10.99	158	2.1295	28.6425	11.8399	21.3847	27.0508	497.5355
0.3292	11.97	172	2.2441	31.8393	13.1308	22.135	30.5866	478.8107
0.2812	12.94	186	2.3464	34.4102	14.3607	23.8634	32.9732	429.9911
0.2443	13.98	201	2.2003	34.8239	14.8042	25.2438	33.0469	392.5385
0.1958	14.61	210	2.5840	29.7482	12.0072	21.348	28.5849	503.5769

Framework versions

• Transformers 4.38.1
• Pytorch 2.2.1+cu121
• Datasets 2.17.1
• Tokenizers 0.15.2