mt5-small-nepali-v2-final

This model is a fine-tuned version of aku47/mt5-small-nepali-v2 on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 1.8995
• Rouge1: 0.1439
• Rouge2: 0.0503
• Rougel: 0.1248
• Rougelsum: 0.1247
• Bertscore Precision: 0.7489
• Bertscore Recall: 0.6377
• Bertscore F1: 0.6884
• Gen Len: 21.0

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: 1e-05
• train_batch_size: 2
• eval_batch_size: 4
• seed: 42
• gradient_accumulation_steps: 8
• total_train_batch_size: 16
• 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
• lr_scheduler_warmup_steps: 100
• num_epochs: 5

Training results

Training Loss	Epoch	Step	Validation Loss	Rouge1	Rouge2	Rougel	Rougelsum	Bertscore Precision	Bertscore Recall	Bertscore F1	Gen Len
8.8303	1.0	144	3.9836	0.0749	0.0168	0.0638	0.0637	0.6942	0.5839	0.6334	21.0
5.3543	2.0	288	2.0607	0.1119	0.0242	0.095	0.0954	0.7256	0.6183	0.6673	21.0
4.317	3.0	432	1.9344	0.1282	0.039	0.1123	0.1122	0.7455	0.6341	0.6849	21.0
4.0117	4.0	576	1.9088	0.1472	0.0505	0.127	0.1267	0.7502	0.6387	0.6896	21.0
3.6721	5.0	720	1.8995	0.1439	0.0503	0.1248	0.1247	0.7489	0.6377	0.6884	21.0

Framework versions

• Transformers 4.53.0
• Pytorch 2.7.1+cu118
• Datasets 3.6.0
• Tokenizers 0.21.2