mt5-small-gigatrue-layercut-D5

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

• Loss: 2.4092

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: 128
• eval_batch_size: 128
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 3

Training results

Training Loss	Epoch	Step	Validation Loss
3.6665	0.1015	3000	2.5503
3.1237	0.2030	6000	2.4814
3.0657	0.3044	9000	2.4631
3.0389	0.4059	12000	2.4435
3.0188	0.5074	15000	2.4455
3.0112	0.6089	18000	2.4229
3.0059	0.7104	21000	2.4302
3.0001	0.8119	24000	2.4221
2.994	0.9133	27000	2.4214
2.9932	1.0148	30000	2.4205
2.991	1.1163	33000	2.4148
2.9857	1.2178	36000	2.4131
2.985	1.3193	39000	2.4148
2.9831	1.4207	42000	2.4104
2.9842	1.5222	45000	2.4128
2.9785	1.6237	48000	2.4131
2.9817	1.7252	51000	2.4099
2.9754	1.8267	54000	2.4114
2.977	1.9282	57000	2.4088
2.9784	2.0296	60000	2.4082
2.9792	2.1311	63000	2.4095
2.9768	2.2326	66000	2.4102
2.9773	2.3341	69000	2.4096
2.9764	2.4356	72000	2.4085
2.9771	2.5370	75000	2.4076
2.9795	2.6385	78000	2.4085
2.9768	2.7400	81000	2.4088
2.9762	2.8415	84000	2.4093
2.9776	2.9430	87000	2.4092

Framework versions

• Transformers 4.45.2
• Pytorch 2.5.1
• Datasets 3.2.0
• Tokenizers 0.20.3