deberta-v3-large-survey-main_passage_consistency-rater-gpt4

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: 0.3012
• Krippendorff: 0.8426
• Spearman: 0.8656
• Absolute Agreement: 0.9183
• Agreement Within One: 0.9673

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	50	2.0918	-0.8602	nan	0.0139	1.0
No log	2.0	100	2.0792	-0.8602	nan	0.0139	1.0
No log	3.0	150	1.8791	-0.0618	-0.1346	0.25	0.8472
No log	4.0	200	2.0894	-0.2303	nan	0.375	0.8194
No log	5.0	250	2.4916	-0.2303	nan	0.375	0.8194
No log	6.0	300	2.6486	-0.2303	nan	0.375	0.8194
No log	7.0	350	2.9221	-0.2303	nan	0.375	0.8194
No log	8.0	400	2.9376	-0.2303	nan	0.375	0.8194
No log	9.0	450	3.0675	-0.2303	nan	0.375	0.8194
1.0223	10.0	500	3.0166	-0.2303	nan	0.375	0.8194
1.0223	11.0	550	3.3407	-0.2303	nan	0.375	0.8194
1.0223	12.0	600	3.4897	0.1396	0.2611	0.375	0.8472
1.0223	13.0	650	3.8130	0.1396	0.2611	0.375	0.8472
1.0223	14.0	700	4.0549	0.1396	0.2611	0.375	0.8472
1.0223	15.0	750	4.3384	-0.2303	nan	0.375	0.8194
1.0223	16.0	800	4.5561	-0.2303	nan	0.375	0.8194
1.0223	17.0	850	4.4345	-0.2303	nan	0.375	0.8194
1.0223	18.0	900	4.8191	-0.1764	-0.0680	0.3611	0.8194
1.0223	19.0	950	4.8287	-0.2303	nan	0.375	0.8194
0.2678	20.0	1000	4.6614	-0.1528	-0.0976	0.3472	0.8194

Framework versions

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