641532fbec060001007ccb05aa2e3655

This model is a fine-tuned version of facebook/opt-350m on the dair-ai/emotion [split] dataset. It achieves the following results on the evaluation set:

• Loss: 0.2116
• Data Size: 1.0
• Epoch Runtime: 72.8638
• Accuracy: 0.9304
• F1 Macro: 0.8851

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	2.5068	0	4.1187	0.1638	0.1153
No log	1	500	2.1301	0.0078	4.5427	0.3105	0.1599
No log	2	1000	1.3699	0.0156	5.2352	0.5227	0.2695
No log	3	1500	0.7060	0.0312	6.6902	0.7611	0.6499
No log	4	2000	0.5494	0.0625	9.2800	0.8211	0.7814
0.0421	5	2500	0.4027	0.125	13.7066	0.9073	0.8665
0.2735	6	3000	0.2669	0.25	23.7577	0.9098	0.8551
0.0347	7	3500	0.2160	0.5	38.7626	0.9219	0.8801
0.1924	8.0	4000	0.1625	1.0	74.3580	0.9249	0.8823
0.1727	9.0	4500	0.1579	1.0	72.7475	0.9325	0.8856
0.1443	10.0	5000	0.2970	1.0	74.4532	0.9148	0.8730
0.1151	11.0	5500	0.2297	1.0	74.1817	0.9229	0.8857
0.1379	12.0	6000	0.2072	1.0	75.1008	0.9299	0.8745
0.1173	13.0	6500	0.1561	1.0	73.1457	0.9309	0.8790
0.1199	14.0	7000	0.1987	1.0	74.2312	0.9254	0.8684
0.1445	15.0	7500	0.2298	1.0	74.1622	0.9244	0.8831
0.1074	16.0	8000	0.2384	1.0	72.0220	0.9259	0.8859
0.0937	17.0	8500	0.2116	1.0	72.8638	0.9304	0.8851

Framework versions

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