Model save

README.md

@@ -18,12 +18,12 @@ should probably proofread and complete it, then remove this comment. -->
 
 This model is a fine-tuned version of [sberbank-ai/ruRoberta-large](https://huggingface.co/sberbank-ai/ruRoberta-large) on an unknown dataset.
 It achieves the following results on the evaluation set:
-- Loss: 2.3311
-- Accuracy: 0.3980
-- Top 2 Accuracy: 0.5651
-- Top 3 Accuracy: 0.6403
-- Roc Auc: 0.9489
-- F1: 0.3823
+- Loss: 3.3205
+- Accuracy: 0.3316
+- Top 2 Accuracy: 0.4452
+- Top 3 Accuracy: 0.5217
+- Roc Auc: 0.9245
+- F1: 0.2788
 
 ## Model description
 
@@ -42,23 +42,24 @@ More information needed
 ### Training hyperparameters
 
 The following hyperparameters were used during training:
-- learning_rate: 5e-05
-- train_batch_size: 8
-- eval_batch_size: 8
+- learning_rate: 1e-05
+- train_batch_size: 12
+- eval_batch_size: 16
 - seed: 42
-- gradient_accumulation_steps: 2
-- total_train_batch_size: 16
 - 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
-- num_epochs: 2
+- lr_scheduler_warmup_ratio: 0.1
+- num_epochs: 4
 - mixed_precision_training: Native AMP
 
 ### Training results
 
 | Training Loss | Epoch | Step | Validation Loss | Accuracy | Top 2 Accuracy | Top 3 Accuracy | Roc Auc | F1     |
 |:-------------:|:-----:|:----:|:---------------:|:--------:|:--------------:|:--------------:|:-------:|:------:|
-| No log        | 1.0   | 196  | 2.6514          | 0.3622   | 0.5102         | 0.5880         | 0.9384  | 0.3265 |
-| No log        | 2.0   | 392  | 2.3311          | 0.3980   | 0.5651         | 0.6403         | 0.9489  | 0.3823 |
+| 4.9101        | 1.0   | 262  | 4.6292          | 0.0816   | 0.1301         | 0.1773         | 0.7633  | 0.0566 |
+| 4.4492        | 2.0   | 524  | 3.8332          | 0.2564   | 0.3737         | 0.4528         | 0.8992  | 0.2009 |
+| 3.9241        | 3.0   | 786  | 3.4448          | 0.3163   | 0.4222         | 0.5089         | 0.9194  | 0.2619 |
+| 3.0492        | 4.0   | 1048 | 3.3205          | 0.3316   | 0.4452         | 0.5217         | 0.9245  | 0.2788 |
 
 
 ### Framework versions