|
--- |
|
license: mit |
|
base_model: cointegrated/rubert-tiny2 |
|
tags: |
|
- generated_from_trainer |
|
metrics: |
|
- accuracy |
|
- f1 |
|
- precision |
|
- recall |
|
model-index: |
|
- name: tiny-rubert |
|
results: [] |
|
--- |
|
|
|
<!-- 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. --> |
|
|
|
# tiny-rubert |
|
|
|
This model is a fine-tuned version of [cointegrated/rubert-tiny2](https://huggingface.co/cointegrated/rubert-tiny2) on an unknown dataset. |
|
It achieves the following results on the evaluation set: |
|
- Loss: 2.5730 |
|
- Accuracy: 0.4956 |
|
- F1: 0.6380 |
|
- Precision: 0.7116 |
|
- Recall: 0.5873 |
|
|
|
## 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 |
|
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08 |
|
- lr_scheduler_type: linear |
|
- lr_scheduler_warmup_steps: 500 |
|
- num_epochs: 10 |
|
|
|
### Training results |
|
|
|
| Training Loss | Epoch | Step | Validation Loss | Accuracy | F1 | Precision | Recall | |
|
|:-------------:|:------:|:-----:|:---------------:|:--------:|:------:|:---------:|:------:| |
|
| No log | 0.2569 | 500 | 4.0706 | 0.0551 | 0.0257 | 0.0397 | 0.0556 | |
|
| 4.3917 | 0.5139 | 1000 | 3.3738 | 0.2871 | 0.2256 | 0.4551 | 0.2810 | |
|
| 2.4111 | 0.7708 | 1500 | 3.0120 | 0.4041 | 0.4675 | 0.6818 | 0.4345 | |
|
| 1.6023 | 1.0277 | 2000 | 2.8194 | 0.4454 | 0.5570 | 0.7232 | 0.4930 | |
|
| 1.2666 | 1.2847 | 2500 | 2.7362 | 0.4553 | 0.5615 | 0.7195 | 0.5 | |
|
| 1.0944 | 1.5416 | 3000 | 2.6636 | 0.4513 | 0.5783 | 0.7227 | 0.5106 | |
|
| 1.0944 | 1.7986 | 3500 | 2.5940 | 0.4543 | 0.5842 | 0.7290 | 0.5134 | |
|
| 1.0351 | 2.0555 | 4000 | 2.5506 | 0.4690 | 0.5953 | 0.7435 | 0.5254 | |
|
| 0.9259 | 2.3124 | 4500 | 2.5396 | 0.4474 | 0.5780 | 0.7272 | 0.5127 | |
|
| 0.802 | 2.5694 | 5000 | 2.4499 | 0.4680 | 0.6044 | 0.7420 | 0.5310 | |
|
| 0.7777 | 2.8263 | 5500 | 2.4295 | 0.4661 | 0.5902 | 0.7239 | 0.5232 | |
|
| 0.7247 | 3.0832 | 6000 | 2.4434 | 0.4631 | 0.5880 | 0.7245 | 0.5197 | |
|
| 0.7247 | 3.3402 | 6500 | 2.4479 | 0.4769 | 0.6023 | 0.7401 | 0.5352 | |
|
| 0.6062 | 3.5971 | 7000 | 2.4713 | 0.4720 | 0.6076 | 0.7465 | 0.5359 | |
|
| 0.6207 | 3.8541 | 7500 | 2.4590 | 0.4779 | 0.6020 | 0.7284 | 0.5359 | |
|
| 0.6021 | 4.1110 | 8000 | 2.4468 | 0.4926 | 0.6333 | 0.7359 | 0.5676 | |
|
| 0.4891 | 4.3679 | 8500 | 2.4930 | 0.4848 | 0.6232 | 0.7313 | 0.5599 | |
|
| 0.4983 | 4.6249 | 9000 | 2.4374 | 0.4936 | 0.6249 | 0.7239 | 0.5676 | |
|
| 0.4983 | 4.8818 | 9500 | 2.4792 | 0.4956 | 0.6246 | 0.7208 | 0.5648 | |
|
| 0.4789 | 5.1387 | 10000 | 2.5257 | 0.4897 | 0.6355 | 0.7117 | 0.5845 | |
|
| 0.4353 | 5.3957 | 10500 | 2.5430 | 0.4946 | 0.6358 | 0.7276 | 0.5761 | |
|
| 0.3995 | 5.6526 | 11000 | 2.5579 | 0.4887 | 0.6340 | 0.7188 | 0.5782 | |
|
| 0.4005 | 5.9096 | 11500 | 2.5249 | 0.4828 | 0.6305 | 0.7014 | 0.5824 | |
|
| 0.3774 | 6.1665 | 12000 | 2.6100 | 0.4838 | 0.6295 | 0.7194 | 0.5725 | |
|
| 0.3774 | 6.4234 | 12500 | 2.5730 | 0.4956 | 0.6380 | 0.7116 | 0.5873 | |
|
| 0.3502 | 6.6804 | 13000 | 2.6117 | 0.4916 | 0.6358 | 0.7066 | 0.5880 | |
|
| 0.3562 | 6.9373 | 13500 | 2.6457 | 0.4956 | 0.6373 | 0.7185 | 0.5838 | |
|
| 0.3453 | 7.1942 | 14000 | 2.6547 | 0.4848 | 0.6316 | 0.7062 | 0.5810 | |
|
| 0.3213 | 7.4512 | 14500 | 2.6828 | 0.4877 | 0.6258 | 0.7035 | 0.5746 | |
|
|
|
|
|
### Framework versions |
|
|
|
- Transformers 4.40.1 |
|
- Pytorch 2.2.1+cu121 |
|
- Tokenizers 0.19.1 |
|
|
