Modelpublic

shanhy/furina_latin_original_amh-esp-eng_train_spearman_corr

sourceHugging Faceupdated 3y agoView on Hugging Face

0likes11downloads

Model Card

furinalatinoriginalamh-esp-engtrainspearmancorr

This model is a fine-tuned version of yihongLiu/furina on the None dataset. It achieves the following results on the evaluation set:

—Loss: 0.0272
—Spearman Corr: 0.7575

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: 2e-05
—trainbatchsize: 32
—evalbatchsize: 128
—seed: 42
—gradientaccumulationsteps: 2
—totaltrainbatch_size: 64
—optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
—lrschedulertype: linear
—num_epochs: 30
—mixedprecisiontraining: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Spearman Corr
No log	1.59	200	0.0206	0.6547
0.0681	3.17	400	0.0237	0.7560
0.0224	4.76	600	0.0168	0.7473
0.0172	6.35	800	0.0199	0.7406
0.0172	7.94	1000	0.0211	0.7572
0.0138	9.52	1200	0.0223	0.7572
0.0115	11.11	1400	0.0240	0.7514
0.0094	12.7	1600	0.0229	0.7535
0.008	14.29	1800	0.0284	0.7531
0.008	15.87	2000	0.0220	0.7526
0.0072	17.46	2200	0.0286	0.7588
0.0065	19.05	2400	0.0284	0.7528
0.0058	20.63	2600	0.0230	0.7549
0.0054	22.22	2800	0.0230	0.7565
0.0054	23.81	3000	0.0244	0.7509
0.0052	25.4	3200	0.0255	0.7557
0.0048	26.98	3400	0.0264	0.7572
0.0046	28.57	3600	0.0272	0.7575

Framework versions

—Transformers 4.37.2
—Pytorch 2.2.0+cu121
—Datasets 2.17.0
—Tokenizers 0.15.2