942bcce4591b5dc9c66e0f01d9185a11

This model is a fine-tuned version of Qwen/Qwen2.5-0.5B on the contemmcm/hate-speech-and-offensive-language dataset. It achieves the following results on the evaluation set:

• Loss: 2.0607
• Data Size: 1.0
• Epoch Runtime: 114.0258
• Accuracy: 0.8918
• F1 Macro: 0.7153

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	50.0901	0	9.7310	0.1591	0.1117
No log	1	619	6.4655	0.0078	10.2054	0.6333	0.3648
No log	2	1238	1.9085	0.0156	11.5071	0.8596	0.6272
0.125	3	1857	2.0719	0.0312	13.9039	0.8856	0.5819
0.125	4	2476	1.4850	0.0625	17.4002	0.8943	0.5950
1.4216	5	3095	1.4310	0.125	24.5541	0.8691	0.6868
0.1121	6	3714	1.2036	0.25	38.7706	0.8965	0.5988
1.1651	7	4333	1.2575	0.5	64.4461	0.8969	0.7464
0.9545	8.0	4952	1.3007	1.0	117.3350	0.9006	0.6336
0.6467	9.0	5571	1.3591	1.0	116.4206	0.8880	0.7136
0.4215	10.0	6190	2.0607	1.0	114.0258	0.8918	0.7153

Framework versions

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