tokipona_model_v0.3

This model is a fine-tuned version of t5-small on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 1.3850
• Bleu: 4.7891
• Gen Len: 17.9507

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: 0.0002
• 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: 15
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Bleu	Gen Len
No log	1.0	55	2.2953	0.285	19.0
No log	2.0	110	1.9083	0.3426	19.0
No log	3.0	165	1.7123	0.6444	18.6404
No log	4.0	220	1.6110	1.1193	17.7291
No log	5.0	275	1.5440	0.9035	17.8621
No log	6.0	330	1.4924	0.8067	17.8424
No log	7.0	385	1.4654	0.8635	17.8079
No log	8.0	440	1.4445	2.3215	17.6059
No log	9.0	495	1.4319	2.5679	17.4384
1.8308	10.0	550	1.4178	2.3622	17.7783
1.8308	11.0	605	1.4011	3.6065	17.6995
1.8308	12.0	660	1.3969	3.8257	17.8768
1.8308	13.0	715	1.3930	4.7373	17.8325
1.8308	14.0	770	1.3864	4.7501	17.9113
1.8308	15.0	825	1.3850	4.7891	17.9507

Framework versions

• Transformers 4.28.0
• Pytorch 2.0.0+cu118
• Datasets 2.11.0
• Tokenizers 0.13.3