File size: 11,142 Bytes
2a56d4f | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 | ---
library_name: transformers
language:
- ru
license: apache-2.0
base_model: deepvk/RuModernBERT-base
tags:
- nli
- russian
- text-classification
- generated_from_trainer
datasets:
- cointegrated/nli-rus-translated-v2021
metrics:
- accuracy
model-index:
- name: rumodernbert-nli
results:
- task:
name: Text Classification
type: text-classification
dataset:
name: cointegrated/nli-rus-translated-v2021
type: cointegrated/nli-rus-translated-v2021
metrics:
- name: Accuracy
type: accuracy
value: 0.8142784874162134
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# rumodernbert-nli
This model is a fine-tuned version of [deepvk/RuModernBERT-base](https://huggingface.co/deepvk/RuModernBERT-base) on the cointegrated/nli-rus-translated-v2021 dataset.
It achieves the following results on the evaluation set:
- Loss: 0.5055
- Accuracy: 0.8143
- Macro F1: 0.8045
- Mean Roc Auc: 0.9351
- Roc Auc Entailment: 0.9496
- Roc Auc Contradiction: 0.9487
- Roc Auc Neutral: 0.9071
## 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: 16
- eval_batch_size: 32
- seed: 42
- distributed_type: multi-GPU
- gradient_accumulation_steps: 2
- total_train_batch_size: 32
- optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
- lr_scheduler_type: cosine
- lr_scheduler_warmup_steps: 0.06
- num_epochs: 3
### Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy | Macro F1 | Mean Roc Auc | Roc Auc Entailment | Roc Auc Contradiction | Roc Auc Neutral |
|:-------------:|:------:|:-----:|:---------------:|:--------:|:--------:|:------------:|:------------------:|:---------------------:|:---------------:|
| 2.2104 | 0.0123 | 500 | 1.0678 | 0.4379 | 0.3603 | 0.5997 | 0.6448 | 0.5677 | 0.5868 |
| 1.5496 | 0.0245 | 1000 | 0.7990 | 0.6590 | 0.6391 | 0.8146 | 0.8588 | 0.8359 | 0.7492 |
| 1.3165 | 0.0368 | 1500 | 0.7175 | 0.7100 | 0.6866 | 0.8577 | 0.8929 | 0.8815 | 0.7985 |
| 1.2338 | 0.0491 | 2000 | 0.6983 | 0.7160 | 0.6927 | 0.8672 | 0.9033 | 0.8898 | 0.8084 |
| 1.1754 | 0.0613 | 2500 | 0.7087 | 0.7241 | 0.7076 | 0.8708 | 0.8986 | 0.9002 | 0.8135 |
| 1.1110 | 0.0736 | 3000 | 0.6735 | 0.7313 | 0.7159 | 0.8796 | 0.9114 | 0.9033 | 0.8242 |
| 1.1363 | 0.0859 | 3500 | 0.6352 | 0.7442 | 0.7293 | 0.8881 | 0.9147 | 0.9095 | 0.8402 |
| 1.1236 | 0.0981 | 4000 | 0.6343 | 0.7435 | 0.7261 | 0.8868 | 0.9141 | 0.9071 | 0.8393 |
| 1.0685 | 0.1104 | 4500 | 0.6201 | 0.7544 | 0.7383 | 0.8960 | 0.9220 | 0.9155 | 0.8505 |
| 1.1451 | 0.1227 | 5000 | 0.6254 | 0.7515 | 0.7284 | 0.8950 | 0.9249 | 0.9150 | 0.8452 |
| 1.0986 | 0.1349 | 5500 | 0.5993 | 0.7610 | 0.7512 | 0.9012 | 0.9275 | 0.9163 | 0.8598 |
| 1.0415 | 0.1472 | 6000 | 0.6263 | 0.7549 | 0.7313 | 0.8992 | 0.9249 | 0.9181 | 0.8548 |
| 1.0306 | 0.1595 | 6500 | 0.6624 | 0.7402 | 0.7137 | 0.8984 | 0.9240 | 0.9190 | 0.8522 |
| 1.0294 | 0.1717 | 7000 | 0.6017 | 0.7693 | 0.7534 | 0.9077 | 0.9289 | 0.9240 | 0.8704 |
| 1.0892 | 0.1840 | 7500 | 0.6379 | 0.7562 | 0.7328 | 0.9042 | 0.9264 | 0.9213 | 0.8648 |
| 1.0483 | 0.1963 | 8000 | 0.5756 | 0.7693 | 0.7520 | 0.9091 | 0.9298 | 0.9266 | 0.8709 |
| 1.0151 | 0.2085 | 8500 | 0.5726 | 0.7742 | 0.7633 | 0.9107 | 0.9313 | 0.9269 | 0.8739 |
| 0.9694 | 0.2208 | 9000 | 0.6317 | 0.7681 | 0.7525 | 0.9087 | 0.9312 | 0.9243 | 0.8706 |
| 1.0105 | 0.2331 | 9500 | 0.5735 | 0.7751 | 0.7643 | 0.9107 | 0.9312 | 0.9246 | 0.8764 |
| 0.9759 | 0.2454 | 10000 | 0.5718 | 0.7778 | 0.7693 | 0.9116 | 0.9352 | 0.9240 | 0.8756 |
| 1.0269 | 0.2576 | 10500 | 0.5584 | 0.7820 | 0.7691 | 0.9146 | 0.9355 | 0.9303 | 0.8782 |
| 1.0052 | 0.2699 | 11000 | 0.5409 | 0.7868 | 0.7762 | 0.9183 | 0.9380 | 0.9336 | 0.8833 |
| 0.9736 | 0.2822 | 11500 | 0.5483 | 0.7864 | 0.7707 | 0.9185 | 0.9377 | 0.9343 | 0.8835 |
| 0.9673 | 0.2944 | 12000 | 0.5449 | 0.7869 | 0.7768 | 0.9189 | 0.9393 | 0.9335 | 0.8840 |
| 0.9870 | 0.3067 | 12500 | 0.5487 | 0.7857 | 0.7737 | 0.9184 | 0.9366 | 0.9345 | 0.8841 |
| 0.9597 | 0.3190 | 13000 | 0.5490 | 0.7883 | 0.7795 | 0.9196 | 0.9370 | 0.9342 | 0.8876 |
| 0.9885 | 0.3312 | 13500 | 0.5532 | 0.7864 | 0.7764 | 0.9203 | 0.9392 | 0.9352 | 0.8865 |
| 0.9791 | 0.3435 | 14000 | 0.5332 | 0.7940 | 0.7853 | 0.9224 | 0.9400 | 0.9377 | 0.8894 |
| 0.9304 | 0.3558 | 14500 | 0.5508 | 0.7930 | 0.7790 | 0.9231 | 0.9400 | 0.9374 | 0.8919 |
| 0.9576 | 0.3680 | 15000 | 0.5403 | 0.7954 | 0.7842 | 0.9241 | 0.9408 | 0.9375 | 0.8940 |
| 0.9085 | 0.3803 | 15500 | 0.5327 | 0.7943 | 0.7860 | 0.9233 | 0.9396 | 0.9396 | 0.8907 |
| 0.9422 | 0.3926 | 16000 | 0.5296 | 0.7986 | 0.7900 | 0.9257 | 0.9432 | 0.9394 | 0.8944 |
| 0.9444 | 0.4048 | 16500 | 0.5108 | 0.8016 | 0.7905 | 0.9265 | 0.9445 | 0.9394 | 0.8957 |
| 0.8995 | 0.4171 | 17000 | 0.5236 | 0.8014 | 0.7932 | 0.9253 | 0.9420 | 0.9412 | 0.8928 |
| 0.9130 | 0.4294 | 17500 | 0.5190 | 0.8010 | 0.7907 | 0.9265 | 0.9427 | 0.9408 | 0.8960 |
| 0.9136 | 0.4416 | 18000 | 0.5214 | 0.8033 | 0.7945 | 0.9277 | 0.9442 | 0.9418 | 0.8971 |
| 0.8579 | 0.4539 | 18500 | 0.5147 | 0.8039 | 0.7940 | 0.9286 | 0.9449 | 0.9414 | 0.8996 |
| 0.8914 | 0.4662 | 19000 | 0.5307 | 0.7969 | 0.7912 | 0.9276 | 0.9431 | 0.9408 | 0.8988 |
| 0.8881 | 0.4784 | 19500 | 0.5401 | 0.7962 | 0.7895 | 0.9236 | 0.9440 | 0.9386 | 0.8883 |
| 0.8726 | 0.4907 | 20000 | 0.5130 | 0.8023 | 0.7956 | 0.9289 | 0.9459 | 0.9410 | 0.8998 |
| 0.8909 | 0.5030 | 20500 | 0.5075 | 0.8038 | 0.7955 | 0.9291 | 0.9454 | 0.9431 | 0.8989 |
| 0.8748 | 0.5152 | 21000 | 0.5080 | 0.8073 | 0.7970 | 0.9301 | 0.9468 | 0.9435 | 0.9000 |
| 0.8879 | 0.5275 | 21500 | 0.5117 | 0.8029 | 0.7952 | 0.9281 | 0.9429 | 0.9426 | 0.8988 |
| 0.8578 | 0.5398 | 22000 | 0.5001 | 0.8081 | 0.7983 | 0.9308 | 0.9469 | 0.9457 | 0.8998 |
| 0.8965 | 0.5520 | 22500 | 0.4930 | 0.8095 | 0.7991 | 0.9315 | 0.9478 | 0.9443 | 0.9023 |
| 0.8542 | 0.5643 | 23000 | 0.5220 | 0.8059 | 0.7968 | 0.9297 | 0.9449 | 0.9428 | 0.9014 |
| 0.8808 | 0.5766 | 23500 | 0.5017 | 0.8060 | 0.7960 | 0.9313 | 0.9468 | 0.9440 | 0.9030 |
| 0.8487 | 0.5888 | 24000 | 0.5146 | 0.8100 | 0.8011 | 0.9317 | 0.9458 | 0.9460 | 0.9035 |
| 0.8643 | 0.6011 | 24500 | 0.5091 | 0.8073 | 0.7981 | 0.9300 | 0.9464 | 0.9420 | 0.9017 |
| 0.8851 | 0.6134 | 25000 | 0.4869 | 0.8136 | 0.8052 | 0.9334 | 0.9483 | 0.9468 | 0.9050 |
| 0.9011 | 0.6256 | 25500 | 0.4874 | 0.8141 | 0.8061 | 0.9343 | 0.9491 | 0.9465 | 0.9072 |
| 0.8509 | 0.6379 | 26000 | 0.4911 | 0.8114 | 0.7984 | 0.9342 | 0.9500 | 0.9474 | 0.9051 |
| 0.8263 | 0.6502 | 26500 | 0.4925 | 0.8150 | 0.8069 | 0.9353 | 0.9502 | 0.9480 | 0.9078 |
| 0.8017 | 0.6624 | 27000 | 0.5023 | 0.8118 | 0.8015 | 0.9331 | 0.9484 | 0.9465 | 0.9043 |
| 0.8267 | 0.6747 | 27500 | 0.5126 | 0.8109 | 0.8040 | 0.9336 | 0.9475 | 0.9472 | 0.9062 |
| 0.8766 | 0.6870 | 28000 | 0.4923 | 0.8141 | 0.8061 | 0.9351 | 0.9494 | 0.9473 | 0.9086 |
| 0.8348 | 0.6992 | 28500 | 0.5225 | 0.8099 | 0.8019 | 0.9295 | 0.9456 | 0.9471 | 0.8957 |
| 0.8355 | 0.7115 | 29000 | 0.5055 | 0.8143 | 0.8045 | 0.9351 | 0.9496 | 0.9487 | 0.9071 |
### Framework versions
- Transformers 5.8.1
- Pytorch 2.11.0+cu130
- Datasets 5.0.0
- Tokenizers 0.22.2
|