bart_finetuned_clarify_aspects

This model is a fine-tuned version of facebook/bart-base on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.0560
• Micro Precision: 0.2456
• Micro Recall: 0.0146
• Micro F1: 0.0275
• Macro Precision: 0.2301
• Macro Recall: 0.0132
• Macro F1: 0.0251
• Bleu: 0.8555
• Rouge1: 0.8184
• Rouge2: 0.5282

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: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• num_epochs: 5
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Micro Precision	Micro Recall	Micro F1	Macro Precision	Macro Recall	Macro F1	Bleu	Rouge1	Rouge2
4.8047	0.2404	50	2.1144	0.1810	0.1332	0.1535	0.0870	0.1541	0.1112	0.6815	0.7396	0.4212
1.7948	0.4808	100	0.8959	0.1842	0.1384	0.1581	0.0894	0.1631	0.1155	0.6860	0.7497	0.4381
0.7301	0.7212	150	0.2402	0.2187	0.0926	0.1301	0.1095	0.0917	0.0998	0.7092	0.7108	0.4486
0.22	0.9615	200	0.0950	0.2541	0.0812	0.1230	0.3683	0.0905	0.1453	0.7932	0.7730	0.4542
0.1005	1.2019	250	0.0699	0.2168	0.1290	0.1618	0.3337	0.1228	0.1796	0.7490	0.7693	0.4552
0.0814	1.4423	300	0.0750	0.1692	0.0468	0.0733	0.1176	0.0407	0.0605	0.8194	0.7906	0.4550
0.075	1.6827	350	0.0681	0.2938	0.0593	0.0987	0.1469	0.0486	0.0731	0.8457	0.8197	0.4467
0.08	1.9231	400	0.0682	0.2427	0.1041	0.1457	0.2116	0.1018	0.1375	0.7777	0.7904	0.4945
0.0712	2.1635	450	0.0682	0.3137	0.0166	0.0316	0.2348	0.0147	0.0277	0.8509	0.7969	0.4711
0.0706	2.4038	500	0.0632	0.3026	0.0239	0.0444	0.2769	0.0211	0.0392	0.8550	0.8220	0.5130
0.0677	2.6442	550	0.0642	0.1622	0.0062	0.0120	0.1548	0.0055	0.0105	0.8520	0.8070	0.4974
0.0664	2.8846	600	0.0604	0.3846	0.0052	0.0103	0.4167	0.0050	0.0098	0.8548	0.8162	0.5217
0.0657	3.125	650	0.0613	0.2763	0.0219	0.0405	0.2929	0.0218	0.0405	0.8583	0.8304	0.5329
0.0634	3.3654	700	0.0608	0.1786	0.0052	0.0101	0.1548	0.0047	0.0092	0.8513	0.8091	0.5055
0.0627	3.6058	750	0.0568	0.3265	0.0166	0.0317	0.3129	0.0147	0.0280	0.8463	0.8097	0.5113
0.0595	3.8462	800	0.0572	0.1	0.0010	0.0021	0.05	0.0007	0.0014	0.8508	0.8198	0.5226
0.0603	4.0865	850	0.0562	0.2381	0.0156	0.0293	0.2455	0.0153	0.0288	0.8528	0.8172	0.5253
0.0588	4.3269	900	0.0565	0.2955	0.0135	0.0259	0.3193	0.0132	0.0253	0.8559	0.8178	0.5243
0.06	4.5673	950	0.0565	0.2931	0.0177	0.0334	0.3056	0.0167	0.0317	0.8559	0.8159	0.5255
0.0614	4.8077	1000	0.0560	0.2456	0.0146	0.0275	0.2301	0.0132	0.0251	0.8555	0.8184	0.5282

Framework versions

• Transformers 4.51.1
• Pytorch 2.5.1+cu124
• Datasets 3.5.0
• Tokenizers 0.21.0