morphoseg-en-byt5

This model is a fine-tuned version of google/byt5-small on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.0142
• Exact Match: 0.9244

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: 0.0003
• train_batch_size: 16
• eval_batch_size: 16
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 32
• optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED 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: 500
• num_epochs: 18

Training results

Training Loss	Epoch	Step	Validation Loss	Exact Match
0.0195	1.0	7646	0.0148	0.8325
0.0139	2.0	15292	0.0115	0.8618
0.0114	3.0	22938	0.0102	0.8764
0.009	4.0	30584	0.0093	0.8874
0.007	5.0	38230	0.0093	0.8965
0.0056	6.0	45876	0.0090	0.9021
0.0049	7.0	53522	0.0093	0.9059
0.004	8.0	61168	0.0095	0.9101
0.0034	9.0	68814	0.0096	0.9112
0.0028	10.0	76460	0.0101	0.9150
0.0023	11.0	84106	0.0108	0.9175
0.0018	12.0	91752	0.0108	0.9185
0.0015	13.0	99398	0.0112	0.9209
0.0011	14.0	107044	0.0121	0.9213
0.001	15.0	114690	0.0123	0.9220
0.0008	16.0	122336	0.0128	0.9237
0.0005	17.0	129982	0.0137	0.9245
0.0005	18.0	137628	0.0142	0.9244

Framework versions

• Transformers 4.57.1
• Pytorch 2.8.0+cu126
• Datasets 4.0.0
• Tokenizers 0.22.1