iitb-en-indic-without-punct

This model is a fine-tuned version of ai4bharat/indictrans2-en-indic-dist-200M on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.3722
• Bleu: 10.1304
• Chrfpp: 32.6831
• Comet: 0.5427
• Bleurt: None
• Gen Len: 20.8656

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 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: 8

Training results

Training Loss	Epoch	Step	Validation Loss	Bleu	Chrfpp	Comet	Bleurt	Gen Len
0.4579	0.5059	6000	0.4194	9.0767	31.3725	0.5333	None	20.8719
0.3759	1.0119	12000	0.4006	9.4117	31.6872	0.5364	None	20.8674
0.3814	1.5178	18000	0.3900	9.3917	31.9292	0.5376	None	20.8716
0.3558	2.0238	24000	0.3832	9.6791	32.0982	0.5386	None	20.8702
0.3387	2.5297	30000	0.3811	9.7865	32.1736	0.5391	None	20.8704
0.3216	3.0357	36000	0.3784	9.895	32.3968	0.54	None	20.8716
0.3256	3.5416	42000	0.3750	9.9879	32.4901	0.5418	None	20.8698
0.2883	4.0476	48000	0.3753	9.9845	32.5095	0.5416	None	20.8671
0.2976	4.5535	54000	0.3730	9.9898	32.4808	0.5414	None	20.8654
0.2744	5.0594	60000	0.3751	9.9484	32.5065	0.5414	None	20.8661
0.2965	5.5654	66000	0.3705	10.0548	32.616	0.5428	None	20.8629
0.2748	6.0713	72000	0.3734	10.0419	32.6004	0.5421	None	20.8667
0.2855	6.5773	78000	0.3718	10.0933	32.6272	0.5422	None	20.8674
0.2725	7.0832	84000	0.3726	10.1122	32.6912	0.5426	None	20.8658
0.2681	7.5892	90000	0.3722	10.1304	32.6831	0.5427	None	20.8656

Framework versions

• Transformers 4.53.2
• Pytorch 2.4.0a0+f70bd71a48.nv24.06
• Datasets 2.21.0
• Tokenizers 0.21.4