hate-phi

This model is a fine-tuned version of microsoft/phi-2 on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.3268

• Classification Report: precision recall f1-score support

       0       0.57      0.08      0.14       438
       1       0.91      0.97      0.93      5755
       2       0.80      0.79      0.80      1242
  

  accuracy 0.89 7435 macro avg 0.76 0.61 0.62 7435

weighted avg 0.87 0.89 0.87 7435

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: 0.0002
• train_batch_size: 64
• eval_batch_size: 16
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 256
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 1
• num_epochs: 1

Training results

Training Loss	Epoch	Step	Validation Loss	Classification Report
0.8106	0.37	25	0.4551	precision recall f1-score support

       0       0.18      0.03      0.04       438
       1       0.85      0.97      0.91      5755
       2       0.75      0.46      0.57      1242

accuracy                           0.83      7435

macro avg 0.59 0.49 0.51 7435 weighted avg 0.79 0.83 0.80 7435 | | 0.3677 | 0.74 | 50 | 0.3374 | precision recall f1-score support

       0       0.51      0.09      0.16       438
       1       0.91      0.95      0.93      5755
       2       0.77      0.83      0.80      1242

accuracy                           0.88      7435

macro avg 0.73 0.63 0.63 7435 weighted avg 0.87 0.88 0.87 7435 |

Framework versions

• PEFT 0.11.1
• Transformers 4.39.3
• Pytorch 2.1.2
• Datasets 2.18.0
• Tokenizers 0.15.2