flan-t5-rouge-durga-q5-clean-4e

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

• Loss: 0.0068
• Rouge1: 0.7279
• Rouge2: 0.6940
• Rougel: 0.7270
• Rougelsum: 0.7260

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: 0.0002
• train_batch_size: 24
• eval_batch_size: 24
• seed: 42
• optimizer: Use 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: 60

Training results

Training Loss	Epoch	Step	Validation Loss	Rouge1	Rouge2	Rougel	Rougelsum
2.1863	1.0	9	1.7470	0.2666	0.0753	0.2615	0.2609
2.2667	2.0	18	1.3513	0.3106	0.1072	0.3001	0.2994
1.4223	3.0	27	1.0986	0.3466	0.1348	0.3382	0.3376
1.5007	4.0	36	0.8739	0.3505	0.1562	0.3430	0.3432
1.4099	5.0	45	0.7260	0.3892	0.1758	0.3780	0.3779
0.9957	6.0	54	0.5489	0.3756	0.1913	0.3661	0.3672
0.9214	7.0	63	0.4321	0.3928	0.2141	0.3832	0.3825
0.7139	8.0	72	0.3473	0.4109	0.2480	0.4020	0.4023
0.6069	9.0	81	0.2602	0.4478	0.2897	0.4377	0.4380
0.5718	10.0	90	0.2200	0.4416	0.2834	0.4325	0.4321
0.4169	11.0	99	0.1599	0.4401	0.3084	0.4346	0.4345
0.3225	12.0	108	0.1387	0.4636	0.3224	0.4579	0.4584
0.3575	13.0	117	0.1143	0.4867	0.3583	0.4806	0.4801
0.2931	14.0	126	0.1022	0.4854	0.3662	0.4820	0.4828
0.2127	15.0	135	0.0913	0.4831	0.3654	0.4798	0.4814
0.2038	16.0	144	0.0848	0.5087	0.3932	0.5069	0.5070
0.264	17.0	153	0.0770	0.5041	0.3951	0.5008	0.5017
0.225	18.0	162	0.0745	0.5207	0.4148	0.5186	0.5191
0.1799	19.0	171	0.0660	0.5238	0.4168	0.5218	0.5214
0.1969	20.0	180	0.0585	0.5367	0.4385	0.5354	0.5348
0.1936	21.0	189	0.0537	0.5336	0.4331	0.5311	0.5313
0.1648	22.0	198	0.0482	0.5679	0.4742	0.5634	0.5651
0.139	23.0	207	0.0474	0.5563	0.4650	0.5535	0.5541
0.1243	24.0	216	0.0424	0.5729	0.4932	0.5733	0.5735
0.1175	25.0	225	0.0405	0.6096	0.5354	0.6074	0.6071
0.1087	26.0	234	0.0374	0.6148	0.5404	0.6140	0.6139
0.08	27.0	243	0.0351	0.6005	0.5205	0.5976	0.6004
0.1133	28.0	252	0.0317	0.5957	0.5140	0.5939	0.5947
0.0514	29.0	261	0.0300	0.6371	0.5702	0.6376	0.6368
0.0819	30.0	270	0.0281	0.6332	0.5680	0.6315	0.6318
0.1079	31.0	279	0.0248	0.6358	0.5696	0.6345	0.6344
0.079	32.0	288	0.0271	0.6231	0.5504	0.6222	0.6222
0.0633	33.0	297	0.0239	0.6607	0.5957	0.6572	0.6584
0.0557	34.0	306	0.0207	0.6658	0.6084	0.6639	0.6636
0.0614	35.0	315	0.0176	0.6767	0.6228	0.6753	0.6744
0.1288	36.0	324	0.0197	0.6603	0.5977	0.6592	0.6583
0.0324	37.0	333	0.0172	0.6853	0.6355	0.6852	0.6849
0.0275	38.0	342	0.0167	0.6881	0.6352	0.6852	0.6853
0.0346	39.0	351	0.0162	0.6772	0.6256	0.6760	0.6759
0.0343	40.0	360	0.0148	0.6960	0.6493	0.6935	0.6937
0.0413	41.0	369	0.0133	0.6883	0.6375	0.6869	0.6862
0.0871	42.0	378	0.0127	0.7170	0.6764	0.7146	0.7148
0.0351	43.0	387	0.0126	0.7072	0.6630	0.7067	0.7066
0.0391	44.0	396	0.0123	0.7108	0.6658	0.7091	0.7095
0.019	45.0	405	0.0122	0.7049	0.6601	0.7045	0.7044
0.0366	46.0	414	0.0117	0.7078	0.6670	0.7077	0.7072
0.0222	47.0	423	0.0110	0.7039	0.6591	0.7033	0.7037
0.03	48.0	432	0.0103	0.7098	0.6643	0.7076	0.7089
0.0534	49.0	441	0.0096	0.7184	0.6801	0.7171	0.7176
0.0363	50.0	450	0.0093	0.7207	0.6829	0.7193	0.7195
0.0175	51.0	459	0.0093	0.7208	0.6849	0.7185	0.7196
0.0325	52.0	468	0.0080	0.7276	0.6930	0.7261	0.7261
0.0341	53.0	477	0.0072	0.7310	0.6980	0.7296	0.7287
0.015	54.0	486	0.0072	0.7261	0.6896	0.7246	0.7244
0.0391	55.0	495	0.0074	0.7279	0.6940	0.7270	0.7260
0.0271	56.0	504	0.0072	0.7251	0.6883	0.7242	0.7240
0.0528	57.0	513	0.0073	0.7243	0.6870	0.7229	0.7226
0.0292	58.0	522	0.0069	0.7279	0.6940	0.7270	0.7260
0.0465	59.0	531	0.0068	0.7279	0.6940	0.7270	0.7260
0.0218	60.0	540	0.0068	0.7279	0.6940	0.7270	0.7260

Framework versions

• Transformers 4.46.0
• Pytorch 2.5.0+cu121
• Datasets 3.0.2
• Tokenizers 0.20.1