391d3d739d4346a0dc59234593c7777a

This model is a fine-tuned version of distilbert/distilroberta-base on the dair-ai/emotion [split] dataset. It achieves the following results on the evaluation set:

• Loss: 0.2241
• Data Size: 1.0
• Epoch Runtime: 21.6721
• Accuracy: 0.9249
• F1 Macro: 0.8811

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	1.8674	0	1.4948	0.0333	0.0107
No log	1	500	1.7166	0.0078	1.7414	0.2908	0.0751
No log	2	1000	1.6041	0.0156	1.8828	0.3488	0.0862
No log	3	1500	1.1701	0.0312	2.4570	0.5867	0.2381
No log	4	2000	0.9732	0.0625	3.1445	0.6764	0.3997
0.0591	5	2500	0.5262	0.125	4.4244	0.8180	0.7584
0.3718	6	3000	0.2670	0.25	7.1239	0.9052	0.8467
0.0363	7	3500	0.2100	0.5	12.0629	0.9219	0.8801
0.1744	8.0	4000	0.1844	1.0	22.2267	0.9289	0.8887
0.143	9.0	4500	0.1592	1.0	21.4378	0.9289	0.8882
0.1431	10.0	5000	0.1762	1.0	21.6793	0.9289	0.8909
0.0985	11.0	5500	0.2915	1.0	21.3642	0.9239	0.8816
0.102	12.0	6000	0.2190	1.0	21.3316	0.9274	0.8864
0.0836	13.0	6500	0.2241	1.0	21.6721	0.9249	0.8811

Framework versions

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