deberta-v3-large-survey-main_passage_old_facts-rater

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: 0.5297
• Krippendorff: 0.9316
• Spearman: 0.9421
• Absolute Agreement: 0.8367
• Agreement Within One: 0.9271

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	2.0081	-0.3387	0.1379	0.125	1.0
No log	2.0	100	1.9950	-0.3387	0.1379	0.125	1.0
No log	3.0	150	1.9832	-0.2518	0.0783	0.125	1.0
No log	4.0	200	1.9482	-0.4725	0.0083	0.1389	0.9861
No log	5.0	250	2.0418	-0.5514	nan	0.1389	1.0
No log	6.0	300	2.1032	-0.5514	nan	0.1389	1.0
No log	7.0	350	2.1151	0.2098	0.1986	0.1806	0.6528
No log	8.0	400	2.1739	0.0597	0.2303	0.1806	0.9028
No log	9.0	450	2.2600	0.0212	0.1446	0.1528	0.875
1.7148	10.0	500	2.1888	0.3454	0.3221	0.25	0.7917
1.7148	11.0	550	2.3269	0.3050	0.3196	0.1944	0.8333
1.7148	12.0	600	2.5569	0.1905	0.2657	0.1806	0.8611
1.7148	13.0	650	2.5631	0.2075	0.1730	0.1667	0.7222
1.7148	14.0	700	2.4919	0.2430	0.2573	0.2222	0.7083
1.7148	15.0	750	2.6314	0.3240	0.3371	0.25	0.7917
1.7148	16.0	800	2.6258	0.2891	0.3225	0.1944	0.6944
1.7148	17.0	850	3.0018	0.3293	0.3565	0.2361	0.8472
1.7148	18.0	900	3.0720	0.2096	0.2075	0.1806	0.7639
1.7148	19.0	950	3.4388	0.2703	0.2818	0.1944	0.7917
0.9808	20.0	1000	3.1775	0.2238	0.2103	0.1528	0.75

Framework versions

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