0aff3102075a38357b52a2e770a9453d

This model is a fine-tuned version of albert/albert-xlarge-v2 on the dair-ai/emotion [split] dataset. It achieves the following results on the evaluation set:

• Loss: 1.5599
• Data Size: 1.0
• Epoch Runtime: 37.0744
• Accuracy: 0.3488
• F1 Macro: 0.0862

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	2.0788	0	1.8951	0.1129	0.0363
No log	1	500	1.5987	0.0078	2.4959	0.3488	0.0862
No log	2	1000	1.5909	0.0156	2.3824	0.3488	0.0862
No log	3	1500	1.5766	0.0312	3.0576	0.3483	0.0866
No log	4	2000	1.6177	0.0625	4.1396	0.2908	0.0751
0.0869	5	2500	1.5779	0.125	6.2636	0.2908	0.0751
1.6051	6	3000	1.5902	0.25	10.7312	0.2908	0.0751
0.2597	7	3500	1.5681	0.5	19.4022	0.3488	0.0862
1.5894	8.0	4000	1.5646	1.0	36.4960	0.3488	0.0862
1.5915	9.0	4500	1.5621	1.0	36.9205	0.3488	0.0862
1.5975	10.0	5000	1.5627	1.0	36.6100	0.3488	0.0862
1.5857	11.0	5500	1.5636	1.0	36.9268	0.3488	0.0862
1.5615	12.0	6000	1.5615	1.0	36.9684	0.3488	0.0862
1.5649	13.0	6500	1.5575	1.0	36.7820	0.3488	0.0862
1.5841	14.0	7000	1.5594	1.0	37.1890	0.3488	0.0862
1.5732	15.0	7500	1.5664	1.0	36.9301	0.2908	0.0751
1.5787	16.0	8000	1.5599	1.0	36.8492	0.3488	0.0862
1.5679	17.0	8500	1.5599	1.0	37.0744	0.3488	0.0862

Framework versions

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