LLama3-2-1B-distortion-fold-3-1a-v1

This model is a fine-tuned version of meta-llama/Llama-3.2-1B on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 6.1273
• Accuracy: 0.3542
• Precision Macro: 0.3359
• Recall Macro: 0.3323
• F1 Macro: 0.3291
• Precision Weighted: 0.3561
• Recall Weighted: 0.3542
• F1 Weighted: 0.3497

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: 2e-05
• train_batch_size: 16
• eval_batch_size: 16
• seed: 42
• optimizer: Use adamw_torch_fused with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.06
• num_epochs: 30
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Precision Macro	Recall Macro	F1 Macro	Precision Weighted	Recall Weighted	F1 Weighted
2.5977	1.0	60	2.0788	0.2884	0.3001	0.2445	0.1977	0.3114	0.2884	0.2302
1.5035	2.0	120	2.0748	0.3323	0.3972	0.2953	0.2875	0.3856	0.3323	0.3133
0.8404	3.0	180	3.9934	0.3354	0.3338	0.2858	0.2729	0.3396	0.3354	0.3010
0.3629	4.0	240	4.9369	0.3229	0.3228	0.2945	0.2686	0.3315	0.3229	0.2949
0.1453	5.0	300	4.6641	0.3229	0.3281	0.2931	0.2883	0.3418	0.3229	0.3140
0.0973	6.0	360	5.6616	0.3292	0.3816	0.3156	0.3132	0.4021	0.3292	0.3316
0.0623	7.0	420	5.7900	0.3354	0.3456	0.3076	0.2996	0.3630	0.3354	0.3260
0.0646	8.0	480	5.4955	0.3542	0.3643	0.3422	0.3359	0.3753	0.3542	0.3486
0.0498	9.0	540	5.7547	0.3417	0.3350	0.3239	0.3159	0.3541	0.3417	0.3328
0.0001	10.0	600	5.9530	0.3605	0.3683	0.3665	0.3487	0.3930	0.3605	0.3575
0.0001	11.0	660	6.3169	0.3574	0.3570	0.3366	0.3368	0.3695	0.3574	0.3548
0.0001	12.0	720	6.1648	0.3542	0.3419	0.3348	0.3304	0.3626	0.3542	0.3486
0.0	13.0	780	6.1273	0.3542	0.3359	0.3323	0.3291	0.3561	0.3542	0.3497

Framework versions

• Transformers 4.57.1
• Pytorch 2.9.1+cu128
• Datasets 4.4.1
• Tokenizers 0.22.1