24e2f10455f6fb7e53c02026d9c88efa

This model is a fine-tuned version of meta-llama/Llama-3.2-1B-Instruct on the contemmcm/hate-speech-and-offensive-language dataset. It achieves the following results on the evaluation set:

• Loss: 2.1551
• Data Size: 1.0
• Epoch Runtime: 85.9187
• Accuracy: 0.8925
• F1 Macro: 0.7099

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	14.0464	0	7.2848	0.1080	0.1330
No log	1	619	3.5370	0.0078	7.2242	0.7407	0.3909
No log	2	1238	2.1501	0.0156	9.6914	0.8531	0.5334
0.0986	3	1857	1.9445	0.0312	11.6966	0.8922	0.5891
0.0986	4	2476	1.3945	0.0625	16.0396	0.8981	0.5994
1.4338	5	3095	1.1395	0.125	20.9956	0.9054	0.7222
0.1157	6	3714	1.2827	0.25	32.2595	0.9058	0.6105
1.2661	7	4333	1.1671	0.5	52.9582	0.8975	0.7445
1.1322	8.0	4952	1.1246	1.0	94.3959	0.9067	0.6603
0.9082	9.0	5571	1.1440	1.0	87.4101	0.9038	0.6965
0.7546	10.0	6190	1.5855	1.0	88.6336	0.9048	0.6894
0.588	11.0	6809	1.6973	1.0	86.6070	0.8931	0.7084
0.5184	12.0	7428	2.1551	1.0	85.9187	0.8925	0.7099

Framework versions

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