66db3780115a74afbf9d80ed99ff7777

This model is a fine-tuned version of mistralai/Mistral-7B-v0.3 on the dair-ai/emotion [split] dataset. It achieves the following results on the evaluation set:

• Loss: 4.5319
• Data Size: 1.0
• Epoch Runtime: 231.3335
• Accuracy: 0.8775
• F1 Macro: 0.8307

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	31.2907	0	6.8668	0.1986	0.1261
No log	1	500	218.9803	0.0078	8.2894	0.3488	0.0862
No log	2	1000	518.5592	0.0156	15.2399	0.0791	0.0259
No log	3	1500	11.6738	0.0312	26.5130	0.1119	0.0335
No log	4	2000	8.1015	0.0625	37.4427	0.2908	0.0751
1.2891	5	2500	8.4414	0.125	58.0272	0.1119	0.0335
7.5373	6	3000	12.7959	0.25	88.5246	0.1361	0.0399
1.1462	7	3500	6.9799	0.5	128.3223	0.3790	0.1195
3.5153	8.0	4000	3.2133	1.0	242.6011	0.7475	0.5815
1.9187	9.0	4500	2.3942	1.0	230.9020	0.8422	0.7131
1.8875	10.0	5000	1.7821	1.0	244.3096	0.8800	0.8168
1.1293	11.0	5500	1.8411	1.0	242.6503	0.8810	0.8363
1.0498	12.0	6000	1.6802	1.0	232.4287	0.8770	0.8209
1.2047	13.0	6500	1.9852	1.0	238.7767	0.8846	0.8287
0.7496	14.0	7000	4.6786	1.0	232.5385	0.8634	0.7718
0.8792	15.0	7500	3.5493	1.0	232.1154	0.8720	0.8183
0.7114	16.0	8000	4.5319	1.0	231.3335	0.8775	0.8307

Framework versions

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