687e517828ebafbeab0f56adcb2f91d8

This model is a fine-tuned version of meta-llama/Llama-3.1-8B on the dair-ai/emotion [split] dataset. It achieves the following results on the evaluation set:

• Loss: 2.4531
• Data Size: 1.0
• Epoch Runtime: 562.5043
• Accuracy: 0.9158
• F1 Macro: 0.8640

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	17.5178	0	7.0141	0.0811	0.0791
No log	1	500	19.0941	0.0078	11.0949	0.2087	0.0995
No log	2	1000	11.1537	0.0156	22.8170	0.2908	0.0751
No log	3	1500	7.6085	0.0312	45.1487	0.3488	0.0862
No log	4	2000	6.3485	0.0625	79.0063	0.3488	0.0862
0.5337	5	2500	6.9886	0.125	115.9590	0.2908	0.0751
6.1782	6	3000	5.0944	0.25	211.3386	0.4748	0.2820
0.4917	7	3500	2.4945	0.5	317.8534	0.8095	0.6641
1.3205	8.0	4000	1.3265	1.0	590.8614	0.8851	0.8442
0.7872	9.0	4500	1.2189	1.0	562.3368	0.9052	0.8589
0.7396	10.0	5000	1.1344	1.0	552.5370	0.9088	0.8625
0.6374	11.0	5500	2.1038	1.0	555.9663	0.8952	0.8525
0.5037	12.0	6000	2.2902	1.0	553.9349	0.8997	0.8348
0.4258	13.0	6500	1.9858	1.0	559.1593	0.9128	0.8663
0.2505	14.0	7000	2.4531	1.0	562.5043	0.9158	0.8640

Framework versions

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