4a4f43e8b968fb3104a0b0b9386fc719

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

• Loss: 0.2233
• Data Size: 1.0
• Epoch Runtime: 52.6587
• Accuracy: 0.9219
• F1 Macro: 0.8850

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	1.8662	0	2.2245	0.0786	0.0245
No log	1	500	1.8208	0.0078	2.9432	0.0444	0.0273
No log	2	1000	1.5772	0.0156	3.5809	0.3488	0.0862
No log	3	1500	1.5099	0.0312	5.1497	0.3690	0.1144
No log	4	2000	1.1891	0.0625	7.1146	0.6210	0.4350
0.0635	5	2500	0.5126	0.125	11.2440	0.8543	0.7894
0.3402	6	3000	0.2690	0.25	17.0273	0.9103	0.8656
0.0416	7	3500	0.2282	0.5	30.0098	0.9168	0.8634
0.2072	8.0	4000	0.2054	1.0	53.2236	0.9209	0.8790
0.1455	9.0	4500	0.1686	1.0	50.6821	0.9229	0.8750
0.1523	10.0	5000	0.1912	1.0	51.8002	0.9264	0.8806
0.1817	11.0	5500	0.2188	1.0	51.2507	0.9279	0.8881
0.1451	12.0	6000	0.2919	1.0	50.9844	0.9234	0.8688
0.1126	13.0	6500	0.2233	1.0	52.6587	0.9219	0.8850

Framework versions

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