README.md

metadata

license: mit
tags:
  - generated_from_trainer
metrics:
  - accuracy
model-index:
  - name: MiniLM-evidence-types
    results: []

MiniLM-evidence-types

This model is a fine-tuned version of microsoft/MiniLM-L12-H384-uncased on the None dataset. It achieves the following results on the evaluation set:

• Loss: 2.3471
• Macro f1: 0.4351
• Weighted f1: 0.7056
• Accuracy: 0.7207
• Balanced accuracy: 0.4063

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: 16
• eval_batch_size: 16
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 20
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Macro f1	Weighted f1	Accuracy	Balanced accuracy
1.249	1.0	250	1.1782	0.2143	0.5844	0.6385	0.2417
1.0481	2.0	500	1.0009	0.3079	0.6757	0.6865	0.3192
0.903	3.0	750	1.0094	0.3105	0.6840	0.6986	0.3179
0.7604	4.0	1000	1.0636	0.3817	0.6834	0.6994	0.3751
0.6367	5.0	1250	1.0813	0.3999	0.6963	0.7108	0.3945
0.5293	6.0	1500	1.1597	0.3909	0.6920	0.6986	0.3895
0.4097	7.0	1750	1.3520	0.3517	0.6739	0.6865	0.3757
0.3442	8.0	2000	1.5343	0.4012	0.6684	0.6743	0.4028
0.2663	9.0	2250	1.5623	0.4241	0.7007	0.7154	0.4052
0.2383	10.0	2500	1.6971	0.4327	0.7080	0.7169	0.4179
0.2053	11.0	2750	1.7675	0.4331	0.7073	0.7177	0.4199
0.1698	12.0	3000	1.8678	0.4381	0.7103	0.7298	0.4097
0.1467	13.0	3250	2.0007	0.4343	0.7113	0.7268	0.4082
0.1098	14.0	3500	2.0797	0.4267	0.7004	0.7131	0.3986
0.1049	15.0	3750	2.2048	0.4190	0.7037	0.7192	0.3939
0.0912	16.0	4000	2.2582	0.4263	0.6903	0.7024	0.4003
0.0678	17.0	4250	2.2735	0.4276	0.7052	0.7222	0.4019
0.0623	18.0	4500	2.3478	0.4317	0.7048	0.7207	0.4030
0.0546	19.0	4750	2.3598	0.4298	0.7043	0.7207	0.4003
0.0415	20.0	5000	2.3471	0.4351	0.7056	0.7207	0.4063

Framework versions

• Transformers 4.19.2
• Pytorch 1.11.0+cu113
• Datasets 2.2.2
• Tokenizers 0.12.1