fa30f88d1f78410d6fce5244df4c541a

This model is a fine-tuned version of meta-llama/Llama-3.1-8B on the contemmcm/hate-speech-and-offensive-language dataset. It achieves the following results on the evaluation set:

• Loss: 1.9789
• Data Size: 1.0
• Epoch Runtime: 762.9467
• Accuracy: 0.8752
• F1 Macro: 0.6641

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	6.5804	0	15.5758	0.4501	0.3084
No log	1	619	167.1974	0.0078	19.7300	0.0601	0.0378
No log	2	1238	2.7409	0.0156	38.3070	0.7672	0.2894
0.2877	3	1857	1.8840	0.0312	60.7182	0.8295	0.5348
0.2877	4	2476	3.7782	0.0625	89.7324	0.1729	0.0984
2.0368	5	3095	1.7158	0.125	134.9102	0.8669	0.6106
0.1384	6	3714	1.8871	0.25	212.5925	0.8821	0.5815
1.3892	7	4333	1.3361	0.5	371.6120	0.8912	0.6110
1.3158	8.0	4952	1.5088	1.0	724.5458	0.8594	0.5627
0.9246	9.0	5571	1.7848	1.0	735.0201	0.8446	0.6739
0.8086	10.0	6190	1.5385	1.0	740.6900	0.8872	0.6511
0.7063	11.0	6809	1.9789	1.0	762.9467	0.8752	0.6641

Framework versions

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