<!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. -->
t5-small-adv-mtop
This model is a fine-tuned version of google/mt5-small on the mtop dataset. It achieves the following results on the evaluation set:
- Loss: 0.1058
- Exact Match: 0.7749
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 |
|---|---|---|---|---|
| 2.4477 | 1.09 | 200 | 0.3427 | 0.2555 |
| 0.7447 | 2.17 | 400 | 0.1441 | 0.5123 |
| 0.4016 | 3.26 | 600 | 0.1122 | 0.5808 |
| 0.4824 | 4.35 | 800 | 0.1117 | 0.5955 |
| 0.4124 | 5.43 | 1000 | 0.1075 | 0.6089 |
| 0.4651 | 6.52 | 1200 | 0.1058 | 0.6094 |
| 0.3521 | 7.61 | 1400 | 0.1114 | 0.6139 |
| 0.3476 | 8.69 | 1600 | 0.1171 | 0.6125 |
| 0.2324 | 9.78 | 1800 | 0.1174 | 0.6161 |
| 0.2406 | 10.87 | 2000 | 0.1227 | 0.6157 |
| 0.1002 | 11.96 | 2200 | 0.1198 | 0.6170 |
| 0.0922 | 13.04 | 2400 | 0.1226 | 0.6134 |
| -0.0436 | 14.13 | 2600 | 0.1182 | 0.6183 |
| -0.1447 | 15.22 | 2800 | 0.1202 | 0.6215 |
| -0.2379 | 16.3 | 3000 | 0.1212 | 0.6206 |
Framework versions
- Transformers 4.24.0
- Pytorch 1.13.0+cu117
- Datasets 2.7.0
- Tokenizers 0.13.2
