praTranv2

<!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. -->

M2M101

This model is a fine-tuned version of facebook/m2m100_418M on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.6766
• Bleu: 15.3416
• Meteor: 0.4723
• Gen Len: 28.0271

Model description

This model translates Maharashtri Prakrit (an ancient Indo-Aryan Language) into English. It is fine-tuned on the M2M100 model.

Intended uses & limitations

The use of this model is educational and is intended for students, linguists and casual learners. This model is currently limited and sometimes could give incorrect translation because of small dataset and low resources ness of Maharashtri Prakrit

Training and evaluation data

This model was trained on Custom made dataset of 1474 Prakrit to English sentences. The evaluation of this model was done using BLEU and METEOR score.

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 1e-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: 20
• mixed_precision_training: Native AMP

Training results

Framework versions

• Transformers 4.45.1
• Pytorch 2.4.0
• Datasets 3.0.1
• Tokenizers 0.20.0