scott156/LongT5BaseNSPCCV1

LongT5-Base-NSPCC

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

• Loss: 1.1612
• Rouge1: 0.4604
• Rouge2: 0.1738
• Rougel: 0.2641
• Rougelsum: 0.2648
• Gen Len: 242.1064

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: 16
• eval_batch_size: 16
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 64
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: cosine
• lr_scheduler_warmup_ratio: 0.03
• num_epochs: 6

Training results

Training Loss	Epoch	Step	Validation Loss	Rouge1	Rouge2	Rougel	Rougelsum	Gen Len
13.1772	1.0	12	2.8918	0.2996	0.0788	0.1529	0.1528	359.617
2.3541	2.0	24	1.2767	0.3935	0.1207	0.1972	0.1965	340.8298
1.5574	3.0	36	1.1901	0.4486	0.1662	0.2444	0.2444	278.3511
1.439	4.0	48	1.1712	0.46	0.1746	0.2628	0.2636	254.266
1.4027	5.0	60	1.1603	0.4625	0.174	0.2619	0.2622	246.0851
1.3858	6.0	72	1.1612	0.4604	0.1738	0.2641	0.2648	242.1064

Framework versions

• Transformers 4.39.2
• Pytorch 2.2.1+cu121
• Datasets 2.18.0
• Tokenizers 0.15.2