b80d142041bc5cd3974a9f327eca895f

This model is a fine-tuned version of facebook/opt-125m on the contemmcm/hate-speech-and-offensive-language dataset. It achieves the following results on the evaluation set:

• Loss: 0.6450
• Data Size: 1.0
• Epoch Runtime: 46.5607
• Accuracy: 0.9000
• F1 Macro: 0.7313

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	0.9260	0	4.5330	0.5528	0.3458
No log	1	619	0.6377	0.0078	5.1011	0.7756	0.3360
No log	2	1238	0.3808	0.0156	5.1247	0.8679	0.5909
0.011	3	1857	0.2926	0.0312	5.9260	0.9032	0.6365
0.011	4	2476	0.3013	0.0625	7.2677	0.9036	0.6150
0.2971	5	3095	0.2990	0.125	9.9859	0.8931	0.7292
0.025	6	3714	0.2720	0.25	15.4999	0.9097	0.6702
0.2554	7	4333	0.2730	0.5	25.5126	0.9067	0.7784
0.222	8.0	4952	0.2676	1.0	47.4744	0.9131	0.7535
0.153	9.0	5571	0.3327	1.0	46.9740	0.9000	0.7278
0.1156	10.0	6190	0.4887	1.0	47.0048	0.9097	0.7234
0.0668	11.0	6809	0.6788	1.0	46.6849	0.8803	0.7303
0.0427	12.0	7428	0.6450	1.0	46.5607	0.9000	0.7313

Framework versions

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