README.md

metadata

license: apache-2.0
tags:
  - generated_from_trainer
metrics:
  - accuracy
  - precision
  - recall
base_model: facebook/bart-base
model-index:
  - name: bart-base-lora
    results: []

bart-base-lora

This model is a fine-tuned version of facebook/bart-base on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.6614
• Accuracy: 0.7909
• Precision: 0.7794
• Recall: 0.7909
• Precision Macro: 0.6664
• Recall Macro: 0.6485
• Macro Fpr: 0.0194
• Weighted Fpr: 0.0186
• Weighted Specificity: 0.9735
• Macro Specificity: 0.9842
• Weighted Sensitivity: 0.7901
• Macro Sensitivity: 0.6485
• F1 Micro: 0.7901
• F1 Macro: 0.6250
• F1 Weighted: 0.7804

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: 8
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 32
• 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	Precision	Recall	Precision Macro	Recall Macro	Macro Fpr	Weighted Fpr	Weighted Specificity	Macro Specificity	Weighted Sensitivity	Macro Sensitivity	F1 Micro	F1 Macro	F1 Weighted
No log	1.0	160	1.3205	0.6112	0.5322	0.6112	0.2887	0.3024	0.0464	0.0435	0.9266	0.9692	0.6112	0.3024	0.6112	0.2871	0.5575
No log	2.0	321	0.8875	0.6995	0.6728	0.6995	0.3822	0.4254	0.0306	0.0298	0.9609	0.9774	0.6995	0.4254	0.6995	0.3948	0.6808
No log	3.0	482	0.8427	0.7064	0.6952	0.7064	0.4131	0.4442	0.0295	0.0288	0.9641	0.9780	0.7064	0.4442	0.7064	0.3969	0.6752
1.2895	4.0	643	0.7719	0.7273	0.7132	0.7273	0.4198	0.4598	0.0264	0.0261	0.9690	0.9798	0.7273	0.4598	0.7273	0.4284	0.7167
1.2895	5.0	803	0.7388	0.7506	0.7400	0.7506	0.5733	0.5165	0.0239	0.0232	0.9697	0.9814	0.7506	0.5165	0.7506	0.5072	0.7368
1.2895	6.0	964	0.7526	0.7444	0.7337	0.7444	0.5703	0.5230	0.0247	0.0239	0.9691	0.9809	0.7444	0.5230	0.7444	0.5088	0.7268
0.7332	7.0	1125	0.7082	0.7552	0.7436	0.7552	0.5665	0.5728	0.0233	0.0226	0.9712	0.9818	0.7552	0.5728	0.7552	0.5609	0.7461
0.7332	8.0	1286	0.7161	0.7583	0.7489	0.7583	0.5641	0.5975	0.0228	0.0223	0.9721	0.9820	0.7583	0.5975	0.7583	0.5756	0.7503
0.7332	9.0	1446	0.6831	0.7777	0.7587	0.7777	0.5781	0.6069	0.0208	0.0200	0.9715	0.9833	0.7777	0.6069	0.7777	0.5875	0.7653
0.6167	10.0	1607	0.6683	0.7862	0.7714	0.7862	0.5917	0.6174	0.0198	0.0191	0.9728	0.9839	0.7862	0.6174	0.7862	0.5987	0.7754
0.6167	11.0	1768	0.6885	0.7761	0.7628	0.7761	0.5817	0.6220	0.0210	0.0202	0.9723	0.9832	0.7761	0.6220	0.7761	0.5946	0.7642
0.6167	12.0	1929	0.6830	0.7870	0.7826	0.7870	0.6627	0.6464	0.0197	0.0190	0.9734	0.9840	0.7870	0.6464	0.7870	0.6214	0.7764
0.5314	13.0	2089	0.6605	0.7916	0.7770	0.7916	0.5965	0.6358	0.0192	0.0185	0.9741	0.9844	0.7916	0.6358	0.7916	0.6111	0.7818
0.5314	14.0	2250	0.6614	0.7909	0.7794	0.7909	0.6368	0.6478	0.0193	0.0185	0.9729	0.9842	0.7909	0.6478	0.7909	0.6261	0.7803
0.5314	14.93	2400	0.6647	0.7901	0.7852	0.7901	0.6664	0.6485	0.0194	0.0186	0.9735	0.9842	0.7901	0.6485	0.7901	0.6250	0.7804

Framework versions

• Transformers 4.35.2
• Pytorch 2.1.0+cu121
• Datasets 2.18.0
• Tokenizers 0.15.1