odinsynth_encoder_decoder_native_hf_test

This model is a fine-tuned version of bert-base-uncased on the enoriega/odinsynth_sequence_dataset synthetic_surface dataset. It achieves the following results on the evaluation set:

• Loss: 0.0533
• Accuracy: 0.9332

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: 5e-05
• train_batch_size: 3
• eval_batch_size: 3
• seed: 42
• gradient_accumulation_steps: 200
• total_train_batch_size: 600
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 20.0

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
6.5753	0.67	60	6.1666	0.0150
2.5262	1.34	120	2.1713	0.9345
0.2343	2.01	180	0.1787	0.9346
0.0713	2.68	240	0.0686	0.9330
0.0631	3.35	300	0.0621	0.9334
0.0603	4.02	360	0.0594	0.9332
0.0589	4.69	420	0.0583	0.9334
0.0579	5.36	480	0.0572	0.9336
0.0575	6.03	540	0.0566	0.9333
0.0561	6.69	600	0.0562	0.9333
0.0559	7.36	660	0.0559	0.9332
0.0551	8.03	720	0.0556	0.9332
0.0548	8.7	780	0.0552	0.9333
0.0546	9.37	840	0.0550	0.9333
0.0539	10.04	900	0.0547	0.9331
0.0546	10.71	960	0.0544	0.9332
0.0538	11.38	1020	0.0543	0.9335
0.0534	12.05	1080	0.0540	0.9333
0.0532	12.72	1140	0.0539	0.9334
0.0525	13.39	1200	0.0538	0.9334
0.0526	14.06	1260	0.0538	0.9331
0.0527	14.73	1320	0.0536	0.9331
0.0529	15.4	1380	0.0536	0.9331
0.0526	16.07	1440	0.0535	0.9331
0.0524	16.74	1500	0.0534	0.9333
0.0516	17.41	1560	0.0534	0.9331
0.0527	18.08	1620	0.0534	0.9332
0.0521	18.74	1680	0.0533	0.9332
0.0519	19.41	1740	0.0533	0.9332

Framework versions

• Transformers 4.27.4
• Pytorch 2.0.0
• Datasets 2.11.0
• Tokenizers 0.11.0