cdp-multi-classifier-weighted

This model is a fine-tuned version of alex-miller/ODABert. It achieves the following results on the evaluation set:

• Loss: 0.8564
• Accuracy: 0.9716
• F1: 0.8484
• Precision: 0.7788
• Recall: 0.9316

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 1e-06
• train_batch_size: 24
• eval_batch_size: 24
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 15

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	F1	Precision	Recall
1.0497	1.0	11302	1.5640	0.9621	0.8011	0.7244	0.8958
0.9103	2.0	22604	1.4417	0.9663	0.8203	0.7522	0.9021
0.7629	3.0	33906	0.9562	0.9661	0.8235	0.7406	0.9272
0.6321	4.0	45208	0.9106	0.9697	0.8376	0.7720	0.9153
0.5464	5.0	56510	0.9811	0.9705	0.8419	0.7760	0.9200
0.5043	6.0	67812	0.9484	0.9700	0.8409	0.7677	0.9296
0.4647	7.0	79114	0.8569	0.9713	0.8465	0.7781	0.9281
0.4215	8.0	90416	0.8620	0.9703	0.8430	0.7682	0.9338
0.3794	9.0	101718	0.8569	0.9704	0.8437	0.7682	0.9357
0.344	10.0	113020	0.8305	0.9708	0.8448	0.7720	0.9328
0.3247	11.0	124322	0.7900	0.9707	0.8446	0.7709	0.9338
0.3159	12.0	135624	0.7838	0.9711	0.8463	0.7734	0.9344
0.3166	13.0	146926	0.8381	0.9710	0.8462	0.7727	0.9351
0.279	14.0	158228	0.8694	0.9718	0.8487	0.7821	0.9277
0.281	15.0	169530	0.8564	0.9716	0.8484	0.7788	0.9316

Framework versions

• Transformers 4.40.1
• Pytorch 2.0.1
• Datasets 2.19.0
• Tokenizers 0.19.1