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: 0.7756
• Rouge1: 0.5243
• Rouge2: 0.242
• Rougel: 0.3113
• Rougelsum: 0.3122
• Gen Len: 331.8511

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: 2
• eval_batch_size: 2
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 8
• 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
4.0417	0.9947	94	0.8455	0.4707	0.1986	0.2704	0.2718	303.4468
1.0117	2.0	189	0.8058	0.5178	0.239	0.3066	0.3077	326.3085
0.886	2.9947	283	0.7798	0.5085	0.2272	0.298	0.2989	348.7979
0.805	4.0	378	0.7725	0.5194	0.2386	0.309	0.31	331.3191
0.7724	4.9947	472	0.7749	0.5224	0.2423	0.3133	0.3147	333.6489
0.7514	5.9683	564	0.7756	0.5243	0.242	0.3113	0.3122	331.8511

Framework versions

• Transformers 4.40.1
• Pytorch 2.2.1+cu121
• Datasets 2.19.0
• Tokenizers 0.19.1