da550b8a9ae61093b57b65d4f74938ef

This model is a fine-tuned version of albert/albert-large-v1 on the contemmcm/hate-speech-and-offensive-language dataset. It achieves the following results on the evaluation set:

• Loss: 0.4673
• Data Size: 1.0
• Epoch Runtime: 40.5796
• Accuracy: 0.8746
• F1 Macro: 0.5671

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: 5e-05
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• distributed_type: multi-GPU
• num_devices: 4
• total_train_batch_size: 32
• total_eval_batch_size: 32
• optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: constant
• num_epochs: 50

Training results

Training Loss	Epoch	Step	Validation Loss	Data Size	Epoch Runtime	Accuracy	F1 Macro
No log	0	0	1.0433	0	3.2013	0.4176	0.2425
No log	1	619	0.6879	0.0078	3.9511	0.7672	0.2894
No log	2	1238	0.6647	0.0156	3.8937	0.7672	0.2894
0.0159	3	1857	0.5087	0.0312	4.4309	0.7881	0.3820
0.0159	4	2476	0.3869	0.0625	5.5436	0.8551	0.5547
0.3881	5	3095	0.4770	0.125	7.7149	0.8107	0.4363
0.038	6	3714	0.3668	0.25	12.0971	0.8817	0.5972
0.43	7	4333	0.4151	0.5	21.6391	0.8764	0.5730
0.4089	8.0	4952	0.4285	1.0	39.3567	0.8876	0.5904
0.5906	9.0	5571	0.6068	1.0	40.4869	0.8086	0.4278
0.4937	10.0	6190	0.4673	1.0	40.5796	0.8746	0.5671

Framework versions

• Transformers 4.57.0
• Pytorch 2.8.0+cu128
• Datasets 4.3.0
• Tokenizers 0.22.1