README.md

metadata

library_name: transformers
base_model: aubmindlab/bert-base-arabertv2
tags:
  - generated_from_trainer
model-index:
  - name: arabert_armis_multitask_hard
    results: []

arabert_armis_multitask_hard

This model is a fine-tuned version of aubmindlab/bert-base-arabertv2 on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.7896
• Disagreement Accuracy: 0.5035
• Disagreement F1: 0.4167
• Target Accuracy: 0.6525
• Target F1: 0.6202

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: 2e-05
• train_batch_size: 32
• eval_batch_size: 32
• seed: 42
• optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• num_epochs: 20

Training results

Training Loss	Epoch	Step	Validation Loss	Disagreement Accuracy	Disagreement F1	Target Accuracy	Target F1
0.8457	1.0	21	0.8400	0.5390	0.4961	0.4610	0.5778
0.8372	2.0	42	0.8325	0.5461	0.4386	0.6241	0.5225
0.8341	3.0	63	0.8290	0.5532	0.4324	0.5390	0.5517
0.8153	4.0	84	0.8240	0.5532	0.496	0.5603	0.5634
0.809	5.0	105	0.8209	0.5319	0.3125	0.5816	0.5874
0.79	6.0	126	0.8150	0.5390	0.4882	0.6099	0.5669
0.7857	7.0	147	0.8119	0.5177	0.3462	0.6312	0.5806
0.7775	8.0	168	0.8090	0.5035	0.3137	0.6099	0.5926
0.7763	9.0	189	0.8041	0.5106	0.448	0.6383	0.5984
0.7685	10.0	210	0.8015	0.5106	0.3670	0.6454	0.6094
0.7652	11.0	231	0.7990	0.5035	0.4262	0.6383	0.6165
0.7597	12.0	252	0.7979	0.5106	0.4651	0.6241	0.6074
0.753	13.0	273	0.7955	0.4894	0.4286	0.6454	0.6212
0.7475	14.0	294	0.7930	0.5035	0.4167	0.6596	0.6190
0.7465	15.0	315	0.7923	0.4965	0.4132	0.6525	0.6202
0.7425	16.0	336	0.7914	0.5035	0.4167	0.6454	0.6032
0.7407	17.0	357	0.7905	0.4965	0.4132	0.6525	0.6202
0.7397	18.0	378	0.7901	0.5248	0.4174	0.6525	0.6202
0.7362	19.0	399	0.7897	0.5035	0.4167	0.6525	0.6202
0.7358	20.0	420	0.7896	0.5035	0.4167	0.6525	0.6202

Framework versions

• Transformers 4.57.1
• Pytorch 2.5.1+cu121
• Datasets 4.3.0
• Tokenizers 0.22.1