1b2c399a575d8605b33f5a8aec8e5731

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

• Loss: 0.2779
• Data Size: 1.0
• Epoch Runtime: 26.7732
• Accuracy: 0.9325
• F1 Macro: 0.8858

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	1.6841	0	1.4950	0.1497	0.0725
No log	1	500	1.5836	0.0078	1.8956	0.3443	0.1353
No log	2	1000	1.5445	0.0156	2.1753	0.3564	0.0957
No log	3	1500	1.2411	0.0312	3.2585	0.5691	0.2313
No log	4	2000	0.8960	0.0625	3.5040	0.7157	0.4885
0.0578	5	2500	0.4283	0.125	5.0210	0.8639	0.8100
0.3167	6	3000	0.2288	0.25	8.0218	0.9189	0.8786
0.0314	7	3500	0.2129	0.5	13.9732	0.9284	0.8892
0.1526	8.0	4000	0.1629	1.0	27.3935	0.9320	0.8911
0.1151	9.0	4500	0.2152	1.0	25.5009	0.9269	0.8793
0.1135	10.0	5000	0.2097	1.0	26.2151	0.9330	0.8922
0.0672	11.0	5500	0.2874	1.0	27.8135	0.9299	0.8895
0.0798	12.0	6000	0.2779	1.0	26.7732	0.9325	0.8858

Framework versions

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