haryoaw's picture
Initial Commit
3f0560b verified
|
raw
history blame
6.45 kB
metadata
library_name: transformers
license: mit
base_model: haryoaw/scenario-TCR-NER_data-univner_half
tags:
  - generated_from_trainer
metrics:
  - precision
  - recall
  - f1
  - accuracy
model-index:
  - name: scenario-non-kd-po-ner-full-xlmr_data-univner_half44
    results: []

scenario-non-kd-po-ner-full-xlmr_data-univner_half44

This model is a fine-tuned version of haryoaw/scenario-TCR-NER_data-univner_half on the None dataset. It achieves the following results on the evaluation set:

  • Loss: 0.1624
  • Precision: 0.8067
  • Recall: 0.8152
  • F1: 0.8109
  • Accuracy: 0.9801

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: 3e-05
  • train_batch_size: 32
  • eval_batch_size: 32
  • seed: 44
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • num_epochs: 30

Training results

Training Loss Epoch Step Validation Loss Precision Recall F1 Accuracy
0.075 0.5828 500 0.0769 0.7627 0.7648 0.7638 0.9772
0.0416 1.1655 1000 0.0813 0.7656 0.8016 0.7832 0.9786
0.0287 1.7483 1500 0.0833 0.7729 0.7989 0.7857 0.9785
0.022 2.3310 2000 0.1011 0.7465 0.7987 0.7717 0.9767
0.0171 2.9138 2500 0.1033 0.7570 0.8166 0.7857 0.9774
0.0133 3.4965 3000 0.0985 0.7734 0.8160 0.7942 0.9786
0.0121 4.0793 3500 0.1039 0.7721 0.8120 0.7916 0.9787
0.0093 4.6620 4000 0.1150 0.7766 0.8031 0.7896 0.9781
0.0083 5.2448 4500 0.1197 0.7771 0.8194 0.7977 0.9788
0.0077 5.8275 5000 0.1180 0.7721 0.8094 0.7903 0.9784
0.0073 6.4103 5500 0.1169 0.7924 0.8061 0.7992 0.9793
0.0065 6.9930 6000 0.1201 0.8011 0.8018 0.8014 0.9791
0.0055 7.5758 6500 0.1152 0.7950 0.7958 0.7954 0.9792
0.0045 8.1585 7000 0.1259 0.7825 0.8133 0.7976 0.9790
0.0039 8.7413 7500 0.1300 0.7878 0.8101 0.7988 0.9793
0.0043 9.3240 8000 0.1279 0.8026 0.7960 0.7993 0.9792
0.0041 9.9068 8500 0.1249 0.8033 0.7997 0.8015 0.9790
0.0033 10.4895 9000 0.1315 0.7890 0.8110 0.7999 0.9793
0.0034 11.0723 9500 0.1268 0.7914 0.8005 0.7959 0.9789
0.0024 11.6550 10000 0.1354 0.7943 0.8088 0.8015 0.9798
0.0023 12.2378 10500 0.1397 0.8003 0.8130 0.8066 0.9798
0.0024 12.8205 11000 0.1428 0.7779 0.8175 0.7972 0.9788
0.0023 13.4033 11500 0.1389 0.7896 0.8070 0.7982 0.9792
0.0022 13.9860 12000 0.1429 0.8 0.8051 0.8025 0.9794
0.0017 14.5688 12500 0.1458 0.7922 0.8113 0.8016 0.9791
0.002 15.1515 13000 0.1430 0.7955 0.8121 0.8037 0.9794
0.0016 15.7343 13500 0.1487 0.7795 0.8185 0.7985 0.9789
0.0019 16.3170 14000 0.1401 0.7994 0.8044 0.8019 0.9796
0.0016 16.8998 14500 0.1468 0.7964 0.8133 0.8048 0.9797
0.0016 17.4825 15000 0.1440 0.7955 0.8100 0.8027 0.9794
0.0012 18.0653 15500 0.1430 0.8002 0.8119 0.8060 0.9798
0.0009 18.6480 16000 0.1558 0.7855 0.8166 0.8007 0.9788
0.0014 19.2308 16500 0.1478 0.8012 0.8119 0.8065 0.9797
0.001 19.8135 17000 0.1498 0.8101 0.8087 0.8094 0.9798
0.0009 20.3963 17500 0.1515 0.7946 0.8116 0.8030 0.9795
0.0008 20.9790 18000 0.1546 0.7967 0.8165 0.8065 0.9796
0.0009 21.5618 18500 0.1553 0.7953 0.8162 0.8056 0.9796
0.0008 22.1445 19000 0.1540 0.8001 0.8103 0.8052 0.9798
0.0006 22.7273 19500 0.1620 0.7877 0.8208 0.8039 0.9793
0.0007 23.3100 20000 0.1605 0.8127 0.7999 0.8062 0.9795
0.0005 23.8928 20500 0.1589 0.7974 0.8132 0.8052 0.9799
0.0005 24.4755 21000 0.1593 0.8016 0.8093 0.8054 0.9797
0.0006 25.0583 21500 0.1561 0.8060 0.8114 0.8087 0.9800
0.0005 25.6410 22000 0.1587 0.8028 0.8152 0.8089 0.9798
0.0004 26.2238 22500 0.1582 0.8058 0.8101 0.8080 0.9802
0.0004 26.8065 23000 0.1612 0.8003 0.8162 0.8081 0.9799
0.0003 27.3893 23500 0.1609 0.8066 0.8142 0.8104 0.9801
0.0003 27.9720 24000 0.1627 0.8032 0.8178 0.8104 0.9801
0.0003 28.5548 24500 0.1616 0.8081 0.8127 0.8104 0.9801
0.0003 29.1375 25000 0.1627 0.8065 0.8153 0.8109 0.9801
0.0002 29.7203 25500 0.1624 0.8067 0.8152 0.8109 0.9801

Framework versions

  • Transformers 4.44.2
  • Pytorch 2.1.1+cu121
  • Datasets 2.14.5
  • Tokenizers 0.19.1