flan-t5-large-invoice-medium

This model is a fine-tuned version of google/flan-t5-large on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.8349
• Rouge1: 19.7835
• Rouge2: 11.9308
• Rougel: 17.7862
• Rougelsum: 19.106
• 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: 5e-05
• train_batch_size: 1
• eval_batch_size: 1
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 20

Training results

Training Loss	Epoch	Step	Validation Loss	Rouge1	Rouge2	Rougel	Rougelsum	Gen Len
1.152	1.0	950	0.6631	17.9812	8.0615	15.6204	17.1827	19.0
0.717	2.0	1900	0.6265	19.7976	11.4928	17.5681	18.5593	19.0
0.6196	3.0	2850	0.5983	18.9511	10.8623	16.7812	17.9926	19.0
0.5513	4.0	3800	0.5915	18.8654	11.6228	17.1068	17.8949	19.0
0.5064	5.0	4750	0.5880	19.031	11.1325	16.8401	17.8006	19.0
0.4634	6.0	5700	0.5857	20.0939	12.744	18.3823	19.164	19.0
0.4224	7.0	6650	0.5931	19.34	11.8669	17.485	18.5155	19.0
0.3967	8.0	7600	0.6193	19.754	11.7313	17.6679	18.8413	19.0
0.3714	9.0	8550	0.6151	19.0538	10.7029	16.7677	18.08	19.0
0.3396	10.0	9500	0.6331	18.718	11.5101	17.2264	17.9486	19.0
0.3089	11.0	10450	0.6602	19.7598	11.5673	17.4428	18.6642	19.0
0.2924	12.0	11400	0.6828	19.3283	11.7675	17.5987	18.5341	19.0
0.2671	13.0	12350	0.7119	19.2428	11.9232	17.6073	18.5031	19.0
0.2573	14.0	13300	0.7336	19.7283	12.3347	17.826	18.8466	19.0
0.2385	15.0	14250	0.7555	19.7352	12.007	17.7395	18.9484	19.0
0.2255	16.0	15200	0.7543	19.479	11.5607	17.492	18.6705	19.0
0.2057	17.0	16150	0.7962	20.0755	12.2036	17.9254	19.2418	19.0
0.2051	18.0	17100	0.7997	19.5207	11.4937	17.319	18.641	19.0
0.1917	19.0	18050	0.8227	20.0937	12.2065	17.787	19.2884	19.0
0.1855	20.0	19000	0.8349	19.7835	11.9308	17.7862	19.106	19.0

Framework versions

• Transformers 4.43.2
• Pytorch 2.2.0a0+81ea7a4
• Datasets 2.20.0
• Tokenizers 0.19.1