my_awesome_billsum_model – AI Model by FolkFoxWalker | AlphaNeural AI

You can deploy this model and start earning money today!

FolkFoxWalker/
my_awesome_billsum_model

like

0

text2text-generation

generated_from_trainer

text-generation-inference

endpoints_compatible

Views

No views yet

my_awesome_billsum_model

This model is a fine-tuned version of t5-small on the billsum dataset. It achieves the following results on the evaluation set:

Loss: 2.5057
Rouge1: 0.1437
Rouge2: 0.0544
Rougel: 0.12
Rougelsum: 0.12
Gen Len: 19.0

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: 2e-05
train_batch_size: 16
eval_batch_size: 16
seed: 42
optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
lr_scheduler_type: linear
num_epochs: 4
mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Rouge1	Rouge2	Rougel	Rougelsum	Gen Len
No log	1.0	62	2.7978	0.1276	0.039	0.1077	0.1076	19.0
No log	2.0	124	2.5889	0.1371	0.0489	0.1153	0.1151	19.0
No log	3.0	186	2.5234	0.1429	0.054	0.1196	0.1194	19.0
No log	4.0	248	2.5057	0.1437	0.0544	0.12	0.12	19.0

Framework versions

Transformers 4.25.1
Pytorch 1.13.0+cu116
Datasets 2.8.0
Tokenizers 0.13.2