scenario-non-kd-po-ner-full-xlmr_data-univner_half66

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: 0.1641
  • Precision: 0.8048
  • Recall: 0.8120
  • F1: 0.8084
  • Accuracy: 0.9797

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: 66
  • 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
0.0757 0.5828 500 0.0751 0.7550 0.7817 0.7681 0.9772
0.0434 1.1655 1000 0.0856 0.7626 0.7987 0.7803 0.9783
0.0293 1.7483 1500 0.0781 0.7835 0.8022 0.7928 0.9792
0.0208 2.3310 2000 0.0929 0.7929 0.7868 0.7898 0.9784
0.0171 2.9138 2500 0.0903 0.7893 0.8176 0.8032 0.9796
0.0125 3.4965 3000 0.1048 0.7915 0.7876 0.7896 0.9779
0.0121 4.0793 3500 0.1080 0.7890 0.8033 0.7961 0.9788
0.0096 4.6620 4000 0.1116 0.7827 0.8083 0.7953 0.9789
0.0083 5.2448 4500 0.1136 0.7957 0.8002 0.7979 0.9786
0.0074 5.8275 5000 0.1162 0.7785 0.8139 0.7958 0.9787
0.0066 6.4103 5500 0.1192 0.7851 0.8058 0.7953 0.9786
0.0065 6.9930 6000 0.1200 0.8077 0.7791 0.7931 0.9787
0.0049 7.5758 6500 0.1227 0.8039 0.7974 0.8007 0.9795
0.0051 8.1585 7000 0.1259 0.7906 0.8058 0.7981 0.9789
0.0041 8.7413 7500 0.1258 0.7992 0.7905 0.7948 0.9792
0.004 9.3240 8000 0.1336 0.7865 0.8072 0.7967 0.9786
0.0039 9.9068 8500 0.1261 0.8145 0.7902 0.8022 0.9792
0.0032 10.4895 9000 0.1309 0.7940 0.7999 0.7970 0.9791
0.0033 11.0723 9500 0.1320 0.8054 0.7869 0.7960 0.9793
0.0026 11.6550 10000 0.1408 0.7915 0.8071 0.7992 0.9789
0.0026 12.2378 10500 0.1404 0.7942 0.8005 0.7973 0.9788
0.0022 12.8205 11000 0.1363 0.7897 0.8134 0.8014 0.9797
0.0024 13.4033 11500 0.1442 0.8065 0.7970 0.8017 0.9793
0.0021 13.9860 12000 0.1401 0.8092 0.7840 0.7964 0.9789
0.0022 14.5688 12500 0.1382 0.8100 0.7983 0.8041 0.9792
0.0015 15.1515 13000 0.1506 0.8066 0.7995 0.8030 0.9793
0.0018 15.7343 13500 0.1437 0.8047 0.7989 0.8018 0.9794
0.002 16.3170 14000 0.1448 0.7997 0.8046 0.8022 0.9794
0.0016 16.8998 14500 0.1466 0.8111 0.7934 0.8021 0.9792
0.0013 17.4825 15000 0.1506 0.8046 0.7943 0.7994 0.9791
0.0012 18.0653 15500 0.1508 0.8038 0.8067 0.8052 0.9796
0.0011 18.6480 16000 0.1493 0.8026 0.8013 0.8020 0.9796
0.0011 19.2308 16500 0.1570 0.7905 0.8048 0.7976 0.9787
0.0011 19.8135 17000 0.1510 0.7980 0.8075 0.8027 0.9793
0.001 20.3963 17500 0.1534 0.7928 0.8173 0.8049 0.9797
0.0009 20.9790 18000 0.1485 0.7916 0.8181 0.8046 0.9798
0.0008 21.5618 18500 0.1509 0.8074 0.8045 0.8060 0.9798
0.0009 22.1445 19000 0.1515 0.8070 0.8084 0.8077 0.9801
0.0006 22.7273 19500 0.1566 0.8022 0.8106 0.8064 0.9798
0.0007 23.3100 20000 0.1620 0.8076 0.7976 0.8026 0.9794
0.0008 23.8928 20500 0.1570 0.8028 0.8084 0.8056 0.9798
0.0005 24.4755 21000 0.1563 0.8020 0.8110 0.8065 0.9798
0.0004 25.0583 21500 0.1610 0.8059 0.8013 0.8036 0.9794
0.0004 25.6410 22000 0.1645 0.8133 0.7943 0.8036 0.9793
0.0004 26.2238 22500 0.1615 0.8031 0.8100 0.8066 0.9798
0.0004 26.8065 23000 0.1630 0.8010 0.8156 0.8083 0.9796
0.0003 27.3893 23500 0.1626 0.8062 0.8114 0.8088 0.9800
0.0003 27.9720 24000 0.1626 0.8054 0.8147 0.8101 0.9800
0.0002 28.5548 24500 0.1639 0.8079 0.8107 0.8093 0.9798
0.0003 29.1375 25000 0.1637 0.8064 0.8085 0.8075 0.9797
0.0002 29.7203 25500 0.1641 0.8048 0.8120 0.8084 0.9797

Framework versions

  • Transformers 4.44.2
  • Pytorch 2.1.1+cu121
  • Datasets 2.14.5
  • Tokenizers 0.19.1
Downloads last month
24
Safetensors
Model size
235M params
Tensor type
F32
·
Inference Examples
This model does not have enough activity to be deployed to Inference API (serverless) yet. Increase its social visibility and check back later, or deploy to Inference Endpoints (dedicated) instead.

Model tree for haryoaw/scenario-non-kd-po-ner-full-xlmr_data-univner_half66