bf09ed6dcff722bf2b60a7c4160452d5

This model is a fine-tuned version of mistralai/Mistral-7B-v0.3 on the contemmcm/hate-speech-and-offensive-language dataset. It achieves the following results on the evaluation set:

• Loss: 3.0826
• Data Size: 1.0
• Epoch Runtime: 295.5108
• Accuracy: 0.8730
• F1 Macro: 0.6522

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	13.1030	0	15.3747	0.5353	0.2975
No log	1	619	8.6099	0.0078	17.4433	0.7672	0.2894
No log	2	1238	181.5505	0.0156	28.5528	0.7672	0.2894
1.8203	3	1857	8.5407	0.0312	41.2820	0.7672	0.2894
1.8203	4	2476	2.7084	0.0625	55.6403	0.7672	0.2894
3.9572	5	3095	2.7772	0.125	82.1349	0.7672	0.2894
0.32	6	3714	2.7825	0.25	102.2462	0.7672	0.2894
3.2022	7	4333	3.2314	0.5	169.1481	0.7672	0.2894
2.0776	8.0	4952	2.5859	1.0	306.8362	0.8186	0.5272
1.5467	9.0	5571	2.0698	1.0	295.9400	0.8245	0.6552
1.4369	10.0	6190	1.5247	1.0	295.4893	0.8841	0.5966
1.5503	11.0	6809	2.3312	1.0	295.1347	0.8744	0.5665
1.2179	12.0	7428	2.1589	1.0	295.5937	0.8817	0.6227
0.6858	13.0	8047	4.0356	1.0	295.6142	0.8776	0.6510
0.9345	14.0	8666	3.0826	1.0	295.5108	0.8730	0.6522

Framework versions

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