Modelpublic

Glow-01/finetuned_pegasus_custom

sourceHugging Faceupdated 3y agoView on Hugging Face

0likes13downloads

Model Card

finetunedpegasuscustom

This model is a fine-tuned version of google/pegasus-multi_news on the None dataset. It achieves the following results on the evaluation set:

—Loss: 1.5792
—Rouge1: 43.3499
—Rouge2: 19.473
—Rougel: 28.3372
—Rougelsum: 39.6698
—Gen Len: 167.2593

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
—trainbatchsize: 1
—evalbatchsize: 1
—seed: 42
—gradientaccumulationsteps: 2
—totaltrainbatch_size: 2
—optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
—lrschedulertype: linear
—num_epochs: 10

Training results

Training Loss	Epoch	Step	Validation Loss	Rouge1	Rouge2	Rougel	Rougelsum	Gen Len
No log	0.98	31	1.6733	45.2908	20.6545	28.8174	41.1913	157.2593
No log	2.0	63	1.6448	45.8258	20.4208	29.3649	41.4304	164.7778
No log	2.98	94	1.6308	45.6111	20.1988	28.7912	41.5061	157.8519
No log	4.0	126	1.6105	45.2388	20.9335	28.8736	41.3696	160.6667
No log	4.98	157	1.6009	44.84	20.5064	29.3276	40.9796	154.0741
No log	6.0	189	1.5903	44.3777	19.987	29.5859	40.7764	163.1111
No log	6.98	220	1.5844	44.3786	20.2566	29.1194	40.9269	160.1111
No log	8.0	252	1.5821	43.3413	19.3	28.3204	39.619	153.6667
No log	8.98	283	1.5796	42.9684	18.9515	27.909	39.166	162.1852
No log	9.84	310	1.5792	43.3499	19.473	28.3372	39.6698	167.2593

Framework versions

—Transformers 4.37.0
—Pytorch 2.1.2
—Datasets 2.1.0
—Tokenizers 0.15.1