deberta-v3-large-survey-main_passage_consistency-rater-all

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: 0.2997
• Krippendorff: 0.8410
• Spearman: 0.8836
• Absolute Agreement: 0.9159
• Agreement Within One: 0.9689

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: 6e-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: 1000
• num_epochs: 20
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Krippendorff	Spearman	Absolute Agreement	Agreement Within One
No log	1.0	55	1.8615	-0.1925	nan	0.0833	0.9167
No log	2.0	110	1.8468	-0.1925	nan	0.0833	0.9167
No log	3.0	165	1.6437	-0.2303	nan	0.375	0.8194
No log	4.0	220	1.7843	-0.2303	nan	0.375	0.8194
No log	5.0	275	1.7113	-0.2303	nan	0.375	0.8194
No log	6.0	330	1.7359	-0.2303	nan	0.375	0.8194
No log	7.0	385	1.6476	-0.2303	nan	0.375	0.8194
No log	8.0	440	1.7123	-0.2303	nan	0.375	0.8194
No log	9.0	495	1.4061	0.0832	0.5104	0.5139	0.8611
1.2442	10.0	550	1.3195	0.0876	0.3809	0.5694	0.8889
1.2442	11.0	605	1.3376	0.0832	0.5104	0.5139	0.8611
1.2442	12.0	660	1.3155	0.1081	0.5170	0.5556	0.875
1.2442	13.0	715	1.2785	0.0948	0.4189	0.5833	0.875
1.2442	14.0	770	1.2800	0.2312	0.3578	0.5556	0.9167
1.2442	15.0	825	1.2081	0.2878	0.5758	0.5833	0.8889
1.2442	16.0	880	1.1245	0.2891	0.4589	0.5556	0.9028
1.2442	17.0	935	1.1434	0.2492	0.3892	0.5833	0.8889
1.2442	18.0	990	1.0987	0.4771	0.3821	0.5556	0.9444
0.6135	19.0	1045	1.0792	0.2770	0.3877	0.5694	0.8889
0.6135	20.0	1100	0.9862	0.0948	0.4189	0.5833	0.875

Framework versions

• Transformers 4.26.0
• Pytorch 1.13.1
• Datasets 2.10.1
• Tokenizers 0.12.1