iva_mt_wslot-m2m100_418M-en-tr

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

• Loss: 0.0136
• Bleu: 63.3126
• Gen Len: 19.5834

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: 4
• eval_batch_size: 4
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 7
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Bleu	Gen Len
0.0175	1.0	2068	0.0151	59.2285	19.5597
0.0121	2.0	4136	0.0138	60.2539	19.3643
0.0087	3.0	6204	0.0134	61.6109	19.3507
0.0065	4.0	8272	0.0134	61.9941	19.6187
0.0049	5.0	10340	0.0134	63.4822	19.6174
0.0039	6.0	12408	0.0136	62.9517	19.6493
0.0031	7.0	14476	0.0136	63.3126	19.5834

Framework versions

• Transformers 4.28.1
• Pytorch 2.0.1+cu118
• Datasets 2.12.0
• Tokenizers 0.13.3

Citation

If you use this model, please cite the following:

@article{Sowanski2023SlotLI,
  title={Slot Lost in Translation? Not Anymore: A Machine Translation Model for Virtual Assistants with Type-Independent Slot Transfer},
  author={Marcin Sowanski and Artur Janicki},
  journal={2023 30th International Conference on Systems, Signals and Image Processing (IWSSIP)},
  year={2023},
  pages={1-5}
}