metadata

license: mit
base_model: microsoft/mdeberta-v3-base
tags:
  - generated_from_trainer
datasets:
  - massive
metrics:
  - accuracy
  - f1
model-index:
  - name: scenario-MDBT-TCR_data-AmazonScience_massive_all_1_1
    results:
      - task:
          name: Text Classification
          type: text-classification
        dataset:
          name: massive
          type: massive
          config: all_1.1
          split: validation
          args: all_1.1
        metrics:
          - name: Accuracy
            type: accuracy
            value: 0.8643440917174317
          - name: F1
            type: f1
            value: 0.8368032657773605

scenario-MDBT-TCR_data-AmazonScience_massive_all_1_1

This model is a fine-tuned version of microsoft/mdeberta-v3-base on the massive dataset. It achieves the following results on the evaluation set:

Loss: 1.0026
Accuracy: 0.8643
F1: 0.8368

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: 32
eval_batch_size: 64
seed: 66
optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
lr_scheduler_type: linear
num_epochs: 5

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	F1
0.5131	0.27	5000	0.6674	0.8368	0.7780
0.3715	0.53	10000	0.6554	0.8527	0.8145
0.3066	0.8	15000	0.6924	0.8471	0.8103
0.2194	1.07	20000	0.7348	0.8548	0.8238
0.2112	1.34	25000	0.7297	0.8581	0.8288
0.1907	1.6	30000	0.7308	0.8558	0.8288
0.1816	1.87	35000	0.7785	0.8565	0.8281
0.1297	2.14	40000	0.8493	0.8567	0.8278
0.127	2.41	45000	0.8757	0.8576	0.8310
0.1148	2.67	50000	0.8581	0.8577	0.8300
0.1287	2.94	55000	0.8479	0.8597	0.8341
0.0875	3.21	60000	0.8763	0.8656	0.8392
0.0832	3.47	65000	0.9379	0.8620	0.8341
0.0837	3.74	70000	0.9044	0.8625	0.8339
0.0617	4.01	75000	0.9840	0.8618	0.8352
0.0524	4.28	80000	0.9955	0.8639	0.8385
0.0496	4.54	85000	1.0026	0.8643	0.8368

Framework versions

Transformers 4.33.3
Pytorch 2.1.1+cu121
Datasets 2.14.5
Tokenizers 0.13.3