eff7df44180322444d1dac074cc45bea

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

• Loss: 0.3032
• Data Size: 1.0
• Epoch Runtime: 15.0475
• Accuracy: 0.9264
• F1 Macro: 0.8885

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	1.7977	0	1.1622	0.0897	0.0613
No log	1	500	1.6469	0.0078	1.3825	0.3070	0.0919
No log	2	1000	1.5541	0.0156	1.4140	0.3488	0.0862
No log	3	1500	1.1470	0.0312	1.7708	0.5932	0.2429
No log	4	2000	0.7320	0.0625	2.2942	0.7702	0.5721
0.0512	5	2500	0.3735	0.125	2.9494	0.8841	0.8398
0.2916	6	3000	0.2462	0.25	4.4101	0.9153	0.8758
0.0358	7	3500	0.2127	0.5	7.9967	0.9163	0.8758
0.1636	8.0	4000	0.1674	1.0	14.7479	0.9249	0.8870
0.1222	9.0	4500	0.1552	1.0	14.7727	0.9340	0.8927
0.0924	10.0	5000	0.1858	1.0	15.1984	0.9325	0.8911
0.0772	11.0	5500	0.2511	1.0	14.4515	0.9259	0.8852
0.0639	12.0	6000	0.3174	1.0	14.6050	0.9244	0.8737
0.0533	13.0	6500	0.3032	1.0	15.0475	0.9264	0.8885

Framework versions

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