bart-large-cnn-finetuned-scope-summarization-train-test-split

This model is a fine-tuned version of facebook/bart-large-cnn on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 1.1824
• Rouge1: 50.8184
• Rouge2: 30.1612
• Rougel: 36.9904
• Rougelsum: 49.4235

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: 5.6e-05
• train_batch_size: 8
• eval_batch_size: 8
• 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
No log	1.0	34	0.7061	44.9133	26.0507	34.7329	43.2231
0.8927	2.0	68	0.7190	46.3702	26.952	36.2244	44.7608
0.617	3.0	102	0.6815	50.5658	28.4213	36.4503	49.3129
0.617	4.0	136	0.7083	50.7172	28.1621	37.0023	49.3293
0.5169	5.0	170	0.6819	50.0364	27.5729	35.9607	48.0331
0.4614	6.0	204	0.7171	51.0974	29.0043	36.4498	49.6124
0.3995	7.0	238	0.7577	50.8879	29.3509	35.9144	49.4029
0.3995	8.0	272	0.8192	50.332	28.9931	36.458	48.8301
0.3185	9.0	306	0.8635	49.4501	27.6673	35.7586	47.891
0.2627	10.0	340	0.9229	51.0907	30.1515	36.5669	49.6126
0.2627	11.0	374	0.9910	49.6877	29.4712	36.5784	48.6087
0.1952	12.0	408	1.0266	51.3167	30.619	36.5097	49.7864
0.1234	13.0	442	1.0313	49.6518	28.2854	35.9773	48.4327
0.0859	14.0	476	1.0791	50.3139	29.9343	35.8756	49.0233
0.0859	15.0	510	1.1431	50.4173	29.047	36.4338	48.8672
0.0591	16.0	544	1.1455	51.733	30.786	37.4082	50.4284
0.0448	17.0	578	1.1353	49.5046	29.3628	36.2758	48.324
0.0341	18.0	612	1.1619	51.7577	31.1043	37.9571	50.6969
0.0341	19.0	646	1.1748	51.4652	30.6309	36.7434	49.9184
0.0255	20.0	680	1.1824	50.8184	30.1612	36.9904	49.4235

Framework versions

• Transformers 4.38.2
• Pytorch 2.2.1+cu121
• Datasets 2.18.0
• Tokenizers 0.15.2