--- library_name: transformers license: apache-2.0 base_model: x2bee/KoModernBERT-base-mlm-v03-retry-ckp02 tags: - generated_from_trainer model-index: - name: KMB_SimCSE_test results: [] --- # KMB_SimCSE_test This model is a fine-tuned version of [x2bee/KoModernBERT-base-mlm-v03-retry-ckp02](https://huggingface.co/x2bee/KoModernBERT-base-mlm-v03-retry-ckp02) on an unknown dataset. It achieves the following results on the evaluation set: - Loss: 0.0463 - Pearson Cosine: 0.8264 - Spearman Cosine: 0.8281 - Pearson Manhattan: 0.8304 - Spearman Manhattan: 0.8368 - Pearson Euclidean: 0.8298 - Spearman Euclidean: 0.8363 - Pearson Dot: 0.7651 - Spearman Dot: 0.7620 ## 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: 16 - eval_batch_size: 16 - seed: 42 - distributed_type: multi-GPU - gradient_accumulation_steps: 8 - total_train_batch_size: 128 - optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments - lr_scheduler_type: linear - lr_scheduler_warmup_ratio: 0.1 - num_epochs: 1.0 ### Training results | Training Loss | Epoch | Step | Validation Loss | Pearson Cosine | Spearman Cosine | Pearson Manhattan | Spearman Manhattan | Pearson Euclidean | Spearman Euclidean | Pearson Dot | Spearman Dot | |:-------------:|:------:|:----:|:---------------:|:--------------:|:---------------:|:-----------------:|:------------------:|:-----------------:|:------------------:|:-----------:|:------------:| | 0.6339 | 0.0469 | 100 | 0.0942 | 0.7833 | 0.7817 | 0.7822 | 0.7875 | 0.7816 | 0.7865 | 0.7100 | 0.7029 | | 0.448 | 0.0937 | 200 | 0.0808 | 0.7856 | 0.7882 | 0.7887 | 0.7945 | 0.7873 | 0.7930 | 0.7399 | 0.7354 | | 0.427 | 0.1406 | 300 | 0.0699 | 0.8037 | 0.8029 | 0.8021 | 0.8090 | 0.8010 | 0.8077 | 0.7505 | 0.7461 | | 0.4052 | 0.1874 | 400 | 0.0616 | 0.8109 | 0.8096 | 0.8094 | 0.8143 | 0.8080 | 0.8128 | 0.7663 | 0.7620 | | 0.4023 | 0.2343 | 500 | 0.0612 | 0.8109 | 0.8135 | 0.8133 | 0.8187 | 0.8113 | 0.8168 | 0.7739 | 0.7705 | | 0.3754 | 0.2812 | 600 | 0.0595 | 0.8105 | 0.8125 | 0.8101 | 0.8162 | 0.8087 | 0.8146 | 0.7706 | 0.7681 | | 0.3729 | 0.3280 | 700 | 0.0619 | 0.8155 | 0.8176 | 0.8144 | 0.8212 | 0.8128 | 0.8196 | 0.7761 | 0.7736 | | 0.341 | 0.3749 | 800 | 0.0530 | 0.8137 | 0.8155 | 0.8201 | 0.8246 | 0.8190 | 0.8234 | 0.7696 | 0.7663 | | 0.3161 | 0.4217 | 900 | 0.0568 | 0.8162 | 0.8182 | 0.8209 | 0.8262 | 0.8198 | 0.8252 | 0.7660 | 0.7625 | | 0.3122 | 0.4686 | 1000 | 0.0541 | 0.8215 | 0.8236 | 0.8220 | 0.8284 | 0.8207 | 0.8268 | 0.7790 | 0.7745 | | 0.3301 | 0.5155 | 1100 | 0.0617 | 0.8116 | 0.8150 | 0.8177 | 0.8228 | 0.8161 | 0.8212 | 0.7638 | 0.7598 | | 0.3637 | 0.5623 | 1200 | 0.0532 | 0.8108 | 0.8145 | 0.8175 | 0.8222 | 0.8156 | 0.8202 | 0.7681 | 0.7643 | | 0.2885 | 0.6092 | 1300 | 0.0451 | 0.8272 | 0.8278 | 0.8275 | 0.8324 | 0.8268 | 0.8318 | 0.7925 | 0.7888 | | 0.2852 | 0.6560 | 1400 | 0.0473 | 0.8246 | 0.8264 | 0.8228 | 0.8281 | 0.8221 | 0.8275 | 0.7893 | 0.7874 | | 0.3225 | 0.7029 | 1500 | 0.0507 | 0.8259 | 0.8284 | 0.8274 | 0.8335 | 0.8263 | 0.8325 | 0.7737 | 0.7708 | | 0.3201 | 0.7498 | 1600 | 0.0467 | 0.8248 | 0.8268 | 0.8232 | 0.8282 | 0.8222 | 0.8274 | 0.7800 | 0.7772 | | 0.3199 | 0.7966 | 1700 | 0.0511 | 0.8215 | 0.8239 | 0.8266 | 0.8322 | 0.8257 | 0.8308 | 0.7702 | 0.7658 | | 0.2431 | 0.8435 | 1800 | 0.0482 | 0.8271 | 0.8287 | 0.8282 | 0.8333 | 0.8277 | 0.8326 | 0.7791 | 0.7749 | | 0.3051 | 0.8903 | 1900 | 0.0465 | 0.8277 | 0.8295 | 0.8257 | 0.8324 | 0.8249 | 0.8319 | 0.7814 | 0.7782 | | 0.3287 | 0.9372 | 2000 | 0.0551 | 0.8207 | 0.8244 | 0.8238 | 0.8296 | 0.8229 | 0.8287 | 0.7620 | 0.7569 | | 0.2889 | 0.9841 | 2100 | 0.0463 | 0.8264 | 0.8281 | 0.8304 | 0.8368 | 0.8298 | 0.8363 | 0.7651 | 0.7620 | ### Framework versions - Transformers 4.48.0.dev0 - Pytorch 2.5.1+cu121 - Datasets 3.1.0 - Tokenizers 0.21.0