Finetuned-hindi-to-english-V5

This model is a fine-tuned version of Helsinki-NLP/opus-mt-hi-en on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 3.3700
• Rouge1: 46.8571
• Rouge2: 20.4177
• Rougel: 41.8681
• Rougelsum: 41.979
• Gen Len: 25.9805

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: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 10
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Rouge1	Rouge2	Rougel	Rougelsum	Gen Len
3.9003	1.0	500	3.5309	44.7687	19.3162	39.7482	39.8743	47.273
3.3953	2.0	1000	3.4295	45.583	19.7519	40.5147	40.6327	31.2255
3.1149	3.0	1500	3.3829	45.9163	19.8789	41.0229	41.1501	34.532
2.9111	4.0	2000	3.3663	45.9453	19.9925	41.0019	41.1038	32.6625
2.7429	5.0	2500	3.3609	46.2821	20.1008	41.2762	41.389	29.1325
2.6212	6.0	3000	3.3592	46.3719	20.2255	41.2826	41.4093	28.0275
2.5156	7.0	3500	3.3632	46.5546	20.3336	41.5246	41.627	25.3715
2.4429	8.0	4000	3.3646	46.5347	20.2066	41.5536	41.6628	27.2935
2.3909	9.0	4500	3.3689	46.9327	20.3823	41.8767	41.9965	25.751
2.358	10.0	5000	3.3700	46.8571	20.4177	41.8681	41.979	25.9805

Framework versions

• Transformers 4.26.1
• Pytorch 1.13.1+cu116
• Datasets 2.9.0
• Tokenizers 0.13.2