Edit model card

t5-base-pointer-adv-cstop_artificial

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

  • Loss: 0.0728
  • Exact Match: 0.7925

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: 8
  • eval_batch_size: 8
  • seed: 42
  • gradient_accumulation_steps: 64
  • 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
1.7423 12.5 200 0.1173 0.2397
0.3678 25.0 400 0.0728 0.3363
0.3202 37.5 600 0.0879 0.3381
0.3452 50.0 800 0.0908 0.3363
0.3099 62.5 1000 0.1056 0.3435
0.3057 75.0 1200 0.1109 0.3470
0.3045 87.5 1400 0.1273 0.3453
0.3052 100.0 1600 0.1065 0.3417
0.3037 112.5 1800 0.1387 0.3381
0.3036 125.0 2000 0.1421 0.3453
0.3023 137.5 2200 0.1649 0.3399
0.3028 150.0 2400 0.1574 0.3399
0.3025 162.5 2600 0.1563 0.3399
0.3017 175.0 2800 0.1589 0.3399
0.302 187.5 3000 0.1587 0.3417

Framework versions

  • Transformers 4.24.0
  • Pytorch 1.13.0+cu117
  • Datasets 2.7.0
  • Tokenizers 0.13.2
Downloads last month
3
Hosted inference API
This model can be loaded on the Inference API on-demand.