librarian-bot's picture
Librarian Bot: Add base_model information to model
b5a4774
|
raw
history blame
7.14 kB
metadata
language:
  - tr
license: apache-2.0
tags:
  - zero-shot-classification
  - nli
  - pytorch
datasets:
  - nli_tr
metrics:
  - accuracy
pipeline_tag: zero-shot-classification
widget:
  - text: Dolar yükselmeye devam ediyor.
    candidate_labels: ekonomi, siyaset, spor
  - text: Senaryo çok saçmaydı, beğendim diyemem.
    candidate_labels: olumlu, olumsuz
base_model: dbmdz/convbert-base-turkish-mc4-cased

convbert-base-turkish-mc4-cased_allnli_tr

This model is a fine-tuned version of dbmdz/convbert-base-turkish-mc4-cased on the None dataset. It achieves the following results on the evaluation set:

  • Loss: 0.5541
  • Accuracy: 0.8111

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: 2e-05
  • train_batch_size: 32
  • eval_batch_size: 32
  • seed: 42
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • num_epochs: 3

Training results

Training Loss Epoch Step Validation Loss Accuracy
0.7338 0.03 1000 0.6722 0.7236
0.603 0.07 2000 0.6465 0.7399
0.5605 0.1 3000 0.5801 0.7728
0.55 0.14 4000 0.5994 0.7626
0.529 0.17 5000 0.5720 0.7697
0.5196 0.2 6000 0.5692 0.7769
0.5117 0.24 7000 0.5725 0.7785
0.5044 0.27 8000 0.5532 0.7787
0.5016 0.31 9000 0.5546 0.7812
0.5031 0.34 10000 0.5461 0.7870
0.4949 0.37 11000 0.5725 0.7826
0.4894 0.41 12000 0.5419 0.7933
0.4796 0.44 13000 0.5278 0.7914
0.4795 0.48 14000 0.5193 0.7953
0.4713 0.51 15000 0.5534 0.7771
0.4738 0.54 16000 0.5098 0.8039
0.481 0.58 17000 0.5244 0.7958
0.4634 0.61 18000 0.5215 0.7972
0.465 0.65 19000 0.5129 0.7985
0.4624 0.68 20000 0.5062 0.8047
0.4597 0.71 21000 0.5114 0.8029
0.4571 0.75 22000 0.5070 0.8073
0.4602 0.78 23000 0.5115 0.7993
0.4552 0.82 24000 0.5085 0.8052
0.4538 0.85 25000 0.5118 0.7974
0.4517 0.88 26000 0.5036 0.8044
0.4517 0.92 27000 0.4930 0.8062
0.4413 0.95 28000 0.5307 0.7964
0.4483 0.99 29000 0.5195 0.7938
0.4036 1.02 30000 0.5238 0.8029
0.3724 1.05 31000 0.5125 0.8082
0.3777 1.09 32000 0.5099 0.8075
0.3753 1.12 33000 0.5172 0.8053
0.367 1.15 34000 0.5188 0.8053
0.3819 1.19 35000 0.5218 0.8046
0.363 1.22 36000 0.5202 0.7993
0.3794 1.26 37000 0.5240 0.8048
0.3749 1.29 38000 0.5026 0.8054
0.367 1.32 39000 0.5198 0.8075
0.3759 1.36 40000 0.5298 0.7993
0.3701 1.39 41000 0.5072 0.8091
0.3742 1.43 42000 0.5071 0.8098
0.3706 1.46 43000 0.5317 0.8037
0.3716 1.49 44000 0.5034 0.8052
0.3717 1.53 45000 0.5258 0.8012
0.3714 1.56 46000 0.5195 0.8050
0.3781 1.6 47000 0.5004 0.8104
0.3725 1.63 48000 0.5124 0.8113
0.3624 1.66 49000 0.5040 0.8094
0.3657 1.7 50000 0.4979 0.8111
0.3669 1.73 51000 0.4968 0.8100
0.3636 1.77 52000 0.5075 0.8079
0.36 1.8 53000 0.4985 0.8110
0.3624 1.83 54000 0.5125 0.8070
0.366 1.87 55000 0.4918 0.8117
0.3655 1.9 56000 0.5051 0.8109
0.3609 1.94 57000 0.5083 0.8105
0.3672 1.97 58000 0.5129 0.8085
0.3545 2.0 59000 0.5467 0.8109
0.2938 2.04 60000 0.5635 0.8049
0.29 2.07 61000 0.5781 0.8041
0.2992 2.11 62000 0.5470 0.8077
0.2957 2.14 63000 0.5765 0.8073
0.292 2.17 64000 0.5472 0.8106
0.2893 2.21 65000 0.5590 0.8085
0.2883 2.24 66000 0.5535 0.8064
0.2923 2.28 67000 0.5508 0.8095
0.2868 2.31 68000 0.5679 0.8098
0.2892 2.34 69000 0.5660 0.8057
0.292 2.38 70000 0.5494 0.8088
0.286 2.41 71000 0.5653 0.8085
0.2939 2.45 72000 0.5673 0.8070
0.286 2.48 73000 0.5600 0.8092
0.2844 2.51 74000 0.5508 0.8095
0.2913 2.55 75000 0.5645 0.8088
0.2859 2.58 76000 0.5677 0.8095
0.2892 2.62 77000 0.5598 0.8113
0.2898 2.65 78000 0.5618 0.8096
0.2814 2.68 79000 0.5664 0.8103
0.2917 2.72 80000 0.5484 0.8122
0.2907 2.75 81000 0.5522 0.8116
0.2896 2.79 82000 0.5540 0.8093
0.2907 2.82 83000 0.5469 0.8104
0.2882 2.85 84000 0.5471 0.8122
0.2878 2.89 85000 0.5532 0.8108
0.2858 2.92 86000 0.5511 0.8115
0.288 2.96 87000 0.5491 0.8111
0.2834 2.99 88000 0.5541 0.8111

Framework versions

  • Transformers 4.12.3
  • Pytorch 1.10.0+cu102
  • Datasets 1.15.1
  • Tokenizers 0.10.3