scott156/LongT5-Large-NSPCC-V2

LongT5-Large-NSPCC

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

• Loss: 1.3710
• Rouge1: 0.4978
• Rouge2: 0.2091
• Rougel: 0.2874
• Rougelsum: 0.2871
• Gen Len: 251.3511

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: 0.0003
• train_batch_size: 1
• eval_batch_size: 1
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 4
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: cosine
• lr_scheduler_warmup_ratio: 0.03
• num_epochs: 8

Training results

Training Loss	Epoch	Step	Validation Loss	Rouge1	Rouge2	Rougel	Rougelsum	Gen Len
6.0401	0.9960	188	2.7089	0.2766	0.0617	0.1655	0.1657	151.7021
2.4805	1.9974	377	1.8809	0.382	0.1178	0.2092	0.2092	211.1809
1.8093	2.9987	566	1.5769	0.4356	0.1527	0.2409	0.2409	246.1277
1.4653	4.0	755	1.4359	0.4661	0.1722	0.26	0.2603	245.0851
1.2626	4.9960	943	1.3908	0.4829	0.1931	0.2717	0.2717	239.8617
1.117	5.9974	1132	1.3724	0.4864	0.1988	0.2804	0.2804	244.4255
1.0404	6.9987	1321	1.3714	0.4914	0.2007	0.2826	0.2821	248.6915
1.0065	7.9682	1504	1.3710	0.4978	0.2091	0.2874	0.2871	251.3511

Framework versions

• Transformers 4.40.2
• Pytorch 2.2.1+cu121
• Datasets 2.19.1
• Tokenizers 0.19.1