furina_afr_corr_2e-05

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

• Loss: 0.0229
• Spearman Corr: 0.7729

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
• train_batch_size: 32
• eval_batch_size: 128
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 64
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 30
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Spearman Corr
No log	0.85	200	0.0390	0.6909
No log	1.69	400	0.0283	0.7242
0.0461	2.54	600	0.0245	0.7522
0.0461	3.38	800	0.0219	0.7600
0.0201	4.23	1000	0.0258	0.7668
0.0201	5.07	1200	0.0234	0.7754
0.0201	5.92	1400	0.0250	0.7778
0.0147	6.77	1600	0.0247	0.7670
0.0147	7.61	1800	0.0236	0.7767
0.0112	8.46	2000	0.0213	0.7708
0.0112	9.3	2200	0.0241	0.7825
0.0092	10.15	2400	0.0228	0.7786
0.0092	10.99	2600	0.0220	0.7789
0.0092	11.84	2800	0.0225	0.7742
0.0073	12.68	3000	0.0229	0.7742
0.0073	13.53	3200	0.0241	0.7745
0.0062	14.38	3400	0.0259	0.7771
0.0062	15.22	3600	0.0254	0.7768
0.0055	16.07	3800	0.0229	0.7729

Framework versions

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