metacognitive-cls

This model is a fine-tuned version of microsoft/deberta-v3-large on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.1024
• Accuracy: 0.9640
• F1: 0.8326
• Precision: 0.8742
• Recall: 0.7947

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: 9.946303722432942e-06
• train_batch_size: 16
• eval_batch_size: 16
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 500
• num_epochs: 12

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	F1	Precision	Recall
0.6685	1.0	76	0.6265	0.7931	0.0543	0.0559	0.0528
0.45	2.0	152	0.2973	0.8983	0.3275	0.6410	0.2199
0.2947	3.0	228	0.2671	0.9069	0.4910	0.6385	0.3988
0.2561	4.0	304	0.2246	0.9234	0.5323	0.8516	0.3871
0.2201	5.0	380	0.1926	0.9442	0.6988	0.8909	0.5748
0.1896	6.0	456	0.1704	0.9439	0.6828	0.9385	0.5367
0.1574	7.0	532	0.1468	0.9515	0.7452	0.9110	0.6305
0.1203	8.0	608	0.1213	0.9591	0.8056	0.8653	0.7537
0.0924	9.0	684	0.1119	0.9634	0.8290	0.8734	0.7889
0.0771	10.0	760	0.1073	0.9620	0.8206	0.8767	0.7713
0.067	11.0	836	0.1016	0.9657	0.8415	0.8762	0.8094
0.0609	12.0	912	0.1024	0.9640	0.8326	0.8742	0.7947

Framework versions

• Transformers 4.40.1
• Pytorch 2.3.0+cu121
• Datasets 2.19.0
• Tokenizers 0.19.1