deberta-v3-large-survey-topicality-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: 1.0013
• Krippendorff: -0.0825
• Spearman: 0.0965
• Absolute Agreement: 0.7324
• Agreement Within One: 0.7663

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	2.0331	-0.7817	nan	0.0417	1.0
No log	2.0	100	2.0130	-0.6164	-0.0607	0.0694	0.9722
No log	3.0	150	1.9945	-0.5886	-0.1318	0.0694	0.9583
No log	4.0	200	1.9436	0.0270	0.1344	0.375	0.8194
No log	5.0	250	2.0502	-0.2309	nan	0.3889	0.7917
No log	6.0	300	1.9935	-0.1983	-0.0674	0.375	0.7917
No log	7.0	350	2.1089	-0.2309	nan	0.3889	0.7917
No log	8.0	400	2.0967	-0.2309	nan	0.3889	0.7917
No log	9.0	450	2.0218	-0.2309	nan	0.3889	0.7917
1.183	10.0	500	2.4122	-0.2309	nan	0.3889	0.7917
1.183	11.0	550	2.1673	-0.2358	-0.1152	0.2361	0.8194
1.183	12.0	600	2.3777	-0.1719	-0.1368	0.2778	0.8056
1.183	13.0	650	2.7792	-0.0861	0.1326	0.3889	0.8056
1.183	14.0	700	2.5054	-0.1648	-0.0297	0.2778	0.8194
1.183	15.0	750	3.0246	-0.0957	-0.0114	0.3611	0.7917
1.183	16.0	800	2.9566	-0.0535	0.0724	0.375	0.8056
1.183	17.0	850	3.1392	-0.0757	0.0155	0.3472	0.8056
1.183	18.0	900	3.0966	-0.1148	-0.0379	0.3194	0.8056
1.183	19.0	950	3.0304	-0.1374	-0.0884	0.3056	0.8056
0.6319	20.0	1000	2.7780	0.0069	0.1298	0.3611	0.8194

Framework versions

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