803c6c2b52bb2545e4ecde0dc3ee0c97

This model is a fine-tuned version of facebook/opt-2.7b on the dair-ai/emotion [split] dataset. It achieves the following results on the evaluation set:

• Loss: 0.2897
• Data Size: 1.0
• Epoch Runtime: 120.7777
• Accuracy: 0.9279
• F1 Macro: 0.8872

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	2.1483	0	5.0103	0.1492	0.1300
No log	1	500	1.9561	0.0078	6.1004	0.3498	0.1415
No log	2	1000	1.3428	0.0156	11.4663	0.5257	0.2679
No log	3	1500	0.8483	0.0312	17.9470	0.7263	0.6509
No log	4	2000	0.7683	0.0625	28.9376	0.7596	0.7318
0.0431	5	2500	0.5398	0.125	38.7031	0.8493	0.7904
0.3929	6	3000	0.5384	0.25	51.2707	0.8357	0.7014
0.0462	7	3500	0.2286	0.5	78.0306	0.9143	0.8669
0.2125	8.0	4000	0.1875	1.0	128.6471	0.9264	0.8736
0.1937	9.0	4500	0.1823	1.0	121.4892	0.9214	0.8768
0.1599	10.0	5000	0.1825	1.0	120.7361	0.9274	0.8864
0.1154	11.0	5500	0.2239	1.0	120.3670	0.9254	0.8884
0.0855	12.0	6000	0.2207	1.0	120.2724	0.9234	0.8858
0.0695	13.0	6500	0.2897	1.0	120.7777	0.9279	0.8872

Framework versions

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