tiny-vanilla-target-glue-stsb

This model is a fine-tuned version of google/bert_uncased_L-2_H-128_A-2 on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.8716
• Pearson: 0.8141
• Spearmanr: 0.8126

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: 3e-05
• train_batch_size: 32
• eval_batch_size: 32
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: constant
• num_epochs: 200

Training results

Training Loss	Epoch	Step	Validation Loss	Pearson	Spearmanr
3.1371	2.78	500	1.1225	0.7236	0.7292
0.9656	5.56	1000	1.0702	0.7703	0.7991
0.7396	8.33	1500	1.0547	0.7931	0.8189
0.6264	11.11	2000	0.9011	0.8123	0.8214
0.5242	13.89	2500	0.8999	0.8135	0.8174
0.4756	16.67	3000	0.9771	0.8142	0.8192
0.4225	19.44	3500	0.9021	0.8168	0.8176
0.3879	22.22	4000	0.9447	0.8176	0.8181
0.3547	25.0	4500	0.8787	0.8226	0.8216
0.3355	27.78	5000	0.9789	0.8157	0.8169
0.3143	30.56	5500	0.9259	0.8152	0.8149
0.2925	33.33	6000	0.8716	0.8141	0.8126

Framework versions

• Transformers 4.26.0.dev0
• Pytorch 1.13.0+cu116
• Datasets 2.8.1.dev0
• Tokenizers 0.13.2