cd40cd2dac776367a2a41fd4af345369

This model is a fine-tuned version of Qwen/Qwen2.5-7B on the contemmcm/hate-speech-and-offensive-language dataset. It achieves the following results on the evaluation set:

• Loss: 1.4832
• Data Size: 1.0
• Epoch Runtime: 527.0791
• Accuracy: 0.8813
• F1 Macro: 0.6131

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	28.6658	0	14.2466	0.0970	0.0785
No log	1	619	10.8858	0.0078	17.7667	0.6575	0.3828
No log	2	1238	4.5616	0.0156	33.1203	0.7746	0.3221
0.2325	3	1857	3.2453	0.0312	56.3642	0.8429	0.5070
0.2325	4	2476	1.2955	0.0625	81.1656	0.8933	0.6557
1.9883	5	3095	2.2069	0.125	129.1398	0.8354	0.6062
0.1438	6	3714	1.7547	0.25	152.9297	0.8289	0.4886
2.9739	7	4333	3.3813	0.5	273.8405	0.7920	0.5871
1.8564	8.0	4952	1.4832	1.0	527.0791	0.8813	0.6131

Framework versions

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