nllb-200-distilled-600M-finetuned_ramayana_sns_prose_lexrank

This model is a fine-tuned version of facebook/nllb-200-distilled-600M on the None dataset. It achieves the following results on the evaluation set:

• Loss: 3.8356
• Rouge1: 15.0234
• Rouge2: 1.2752
• Rougel: 12.4341
• Rougelsum: 13.036

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-06
• train_batch_size: 5
• eval_batch_size: 5
• 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
4.9309	1.0	86	4.4877	17.9643	1.8368	13.627	16.743
4.5531	2.0	172	4.2961	7.7741	0.7819	6.5664	7.0319
4.4082	3.0	258	4.1867	15.3171	1.2616	12.5029	13.4603
4.3142	4.0	344	4.1128	14.7478	1.299	12.4545	12.8832
4.22	5.0	430	4.0577	14.6397	1.1974	12.3644	12.5395
4.1743	6.0	516	4.0173	14.7595	1.3556	12.4877	12.6788
4.1279	7.0	602	3.9858	14.3561	1.3361	12.071	12.5251
4.0927	8.0	688	3.9564	15.0213	1.3697	12.7109	13.1084
4.0625	9.0	774	3.9320	15.2813	1.3317	12.635	13.4154
4.0361	10.0	860	3.9113	15.0786	1.2544	12.6139	13.0141
3.9913	11.0	946	3.8951	15.0242	1.2899	12.7049	13.1678
3.9949	12.0	1032	3.8822	15.1567	1.3332	12.7349	13.1691
3.9643	13.0	1118	3.8724	15.0434	1.2552	12.6509	13.1845
3.96	14.0	1204	3.8608	14.5834	1.2768	12.1898	12.5734
3.9524	15.0	1290	3.8533	14.6872	1.2161	12.2549	12.7557
3.9345	16.0	1376	3.8443	15.0962	1.3235	12.5689	13.1217
3.9359	17.0	1462	3.8407	14.7724	1.2323	12.4059	12.6775
3.9213	18.0	1548	3.8367	14.6599	1.285	12.2237	12.7231
3.9213	19.0	1634	3.8356	15.0234	1.2752	12.4341	13.036

Framework versions

• Transformers 4.41.2
• Pytorch 2.0.1+cu117
• Datasets 2.19.2
• Tokenizers 0.19.1