README.md

metadata

language:
  - en
license: apache-2.0
tags:
  - generated_from_trainer
datasets:
  - glue
metrics:
  - accuracy
  - f1
model-index:
  - name: mobilebert_sa_GLUE_Experiment_logit_kd_pretrain_qqp
    results:
      - task:
          name: Text Classification
          type: text-classification
        dataset:
          name: GLUE QQP
          type: glue
          config: qqp
          split: validation
          args: qqp
        metrics:
          - name: Accuracy
            type: accuracy
            value: 0.9133069502844422
          - name: F1
            type: f1
            value: 0.8843348843348843

mobilebert_sa_GLUE_Experiment_logit_kd_pretrain_qqp

This model is a fine-tuned version of gokuls/mobilebert_sa_pre-training-complete on the GLUE QQP dataset. It achieves the following results on the evaluation set:

• Loss: 0.1715
• Accuracy: 0.9133
• F1: 0.8843
• Combined Score: 0.8988

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: 128
• eval_batch_size: 128
• seed: 10
• distributed_type: multi-GPU
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 50

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	F1	Combined Score
0.3902	1.0	2843	0.2870	0.8788	0.8463	0.8625
0.226	2.0	5686	0.1968	0.8979	0.8649	0.8814
0.1681	3.0	8529	0.1910	0.9038	0.8719	0.8878
0.1346	4.0	11372	0.1989	0.9055	0.8772	0.8914
0.1124	5.0	14215	0.1970	0.9049	0.8747	0.8898
0.0978	6.0	17058	0.1876	0.9095	0.8806	0.8950
0.0876	7.0	19901	0.1893	0.9077	0.8773	0.8925
0.0802	8.0	22744	0.1940	0.9067	0.8773	0.8920
0.0746	9.0	25587	0.1846	0.9090	0.8787	0.8938
0.0699	10.0	28430	0.1890	0.9093	0.8809	0.8951
0.0663	11.0	31273	0.1803	0.9103	0.8804	0.8953
0.0632	12.0	34116	0.1905	0.9084	0.8805	0.8945
0.0606	13.0	36959	0.1835	0.9094	0.8813	0.8953
0.0583	14.0	39802	0.1786	0.9112	0.8805	0.8958
0.0562	15.0	42645	0.1900	0.9091	0.8817	0.8954
0.0546	16.0	45488	0.1753	0.9126	0.8825	0.8975
0.0529	17.0	48331	0.1761	0.9121	0.8825	0.8973
0.0515	18.0	51174	0.1784	0.9129	0.8842	0.8986
0.0501	19.0	54017	0.1730	0.9129	0.8847	0.8988
0.049	20.0	56860	0.1812	0.9116	0.8835	0.8975
0.0479	21.0	59703	0.1751	0.9115	0.8830	0.8972
0.0469	22.0	62546	0.1737	0.9120	0.8833	0.8976
0.0461	23.0	65389	0.1739	0.9129	0.8844	0.8986
0.0452	24.0	68232	0.1715	0.9133	0.8843	0.8988
0.0447	25.0	71075	0.1748	0.9119	0.8844	0.8982
0.0437	26.0	73918	0.1734	0.9129	0.8841	0.8985
0.0431	27.0	76761	0.1727	0.9125	0.8830	0.8977
0.0425	28.0	79604	0.1803	0.9120	0.8851	0.8985
0.0419	29.0	82447	0.1720	0.9124	0.8835	0.8980

Framework versions

• Transformers 4.26.0
• Pytorch 1.14.0a0+410ce96
• Datasets 2.9.0
• Tokenizers 0.13.2