bge-m3-edu-scorer-lr3e5-bs32

This model is a fine-tuned version of BAAI/bge-m3 on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 1.2380
• Precision: 0.4817
• Recall: 0.345
• F1 Macro: 0.3403
• Accuracy: 0.3709

Model description

More information needed

Intended uses & limitations

More information needed

Test results

Binary classification accuracy (threshold at label 3) ≈ 77.45%

Test Report:

              precision    recall  f1-score   support

           0       0.81      0.52      0.63       100
           1       0.36      0.47      0.41       100
           2       0.28      0.52      0.36       100
           3       0.28      0.35      0.31       100
           4       0.42      0.15      0.22       100
           5       0.75      0.06      0.11        50

    accuracy                           0.37       550
   macro avg       0.48      0.34      0.34       550
weighted avg       0.46      0.37      0.36       550

Confusion Matrix:

[[52 39  8  1  0  0]
 [10 47 38  5  0  0]
 [ 2 34 52 11  1  0]
 [ 0  8 49 35  7  1]
 [ 0  3 32 50 15  0]
 [ 0  1 10 23 13  3]]

Test metrics

  epoch                   =       20.0
  eval_accuracy           =     0.3709
  eval_f1_macro           =     0.3403
  eval_loss               =      1.238
  eval_precision          =     0.4817
  eval_recall             =      0.345
  eval_runtime            = 0:00:06.03
  eval_samples_per_second =     91.183
  eval_steps_per_second   =      2.984

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 3e-05
• train_batch_size: 32
• eval_batch_size: 32
• seed: 0
• optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 20

Training results

Framework versions

• Transformers 4.53.2
• Pytorch 2.7.1+cu126
• Datasets 4.0.0
• Tokenizers 0.21.2