t5-small-vanilla-cstop_artificial

This model is a fine-tuned version of google/mt5-small on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.1506
• Exact Match: 0.5725

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: 0.001
• train_batch_size: 16
• eval_batch_size: 16
• seed: 42
• gradient_accumulation_steps: 32
• total_train_batch_size: 512
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• training_steps: 3000

Training results

Training Loss	Epoch	Step	Validation Loss	Exact Match
1.4041	28.5	200	0.1008	0.4758
0.047	57.13	400	0.1029	0.5367
0.021	85.63	600	0.1077	0.5617
0.012	114.25	800	0.1214	0.5689
0.0079	142.75	1000	0.1273	0.5671
0.0809	171.38	1200	0.1192	0.5653
0.0063	199.88	1400	0.1329	0.5653
0.0042	228.5	1600	0.1402	0.5707
0.0036	257.13	1800	0.1335	0.5617
0.0029	285.63	2000	0.1423	0.5689
0.0023	314.25	2200	0.1515	0.5671
0.0019	342.75	2400	0.1569	0.5689
0.0018	371.38	2600	0.1517	0.5689
0.0016	399.88	2800	0.1527	0.5725
0.0016	428.5	3000	0.1506	0.5725

Framework versions

• Transformers 4.24.0
• Pytorch 1.13.0+cu117
• Datasets 2.7.0
• Tokenizers 0.13.2