version 2

README.md

@@ -18,8 +18,8 @@ should probably proofread and complete it, then remove this comment. -->
 
 This model is a fine-tuned version of [facebook/bart-large](https://huggingface.co/facebook/bart-large) on the None dataset.
 It achieves the following results on the evaluation set:
-- Loss: 1.0664
-- Sacrebleu: 66.6476
+- Loss: 0.6032
+- Sacrebleu: 65.7419
 
 ## Model description
 
@@ -39,50 +39,22 @@ More information needed
 
 The following hyperparameters were used during training:
 - learning_rate: 2e-05
-- train_batch_size: 16
-- eval_batch_size: 16
+- train_batch_size: 32
+- eval_batch_size: 32
 - seed: 42
-- gradient_accumulation_steps: 4
-- total_train_batch_size: 64
+- gradient_accumulation_steps: 8
+- total_train_batch_size: 256
 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
 - lr_scheduler_type: linear
-- num_epochs: 30
+- num_epochs: 2
 - mixed_precision_training: Native AMP
 
 ### Training results
 
 | Training Loss | Epoch | Step | Validation Loss | Sacrebleu |
 |:-------------:|:-----:|:----:|:---------------:|:---------:|
-| No log        | 1.0   | 218  | 0.5750          | 66.6584   |
-| No log        | 2.0   | 437  | 0.5581          | 66.9419   |
-| No log        | 3.0   | 656  | 0.5662          | 66.8166   |
-| No log        | 4.0   | 875  | 0.6339          | 66.8911   |
-| No log        | 5.0   | 1093 | 0.6190          | 66.4260   |
-| No log        | 6.0   | 1312 | 0.6760          | 66.7698   |
-| No log        | 7.0   | 1531 | 0.6708          | 66.7328   |
-| No log        | 8.0   | 1750 | 0.7686          | 66.6153   |
-| No log        | 9.0   | 1968 | 0.7157          | 66.7670   |
-| No log        | 10.0  | 2187 | 0.7567          | 66.6510   |
-| No log        | 11.0  | 2406 | 0.7699          | 66.5710   |
-| No log        | 12.0  | 2625 | 0.8145          | 66.7658   |
-| No log        | 13.0  | 2843 | 0.8292          | 66.4557   |
-| No log        | 14.0  | 3062 | 0.8610          | 66.7477   |
-| No log        | 15.0  | 3281 | 0.8962          | 66.4487   |
-| No log        | 16.0  | 3500 | 0.9000          | 66.6798   |
-| No log        | 17.0  | 3718 | 0.9376          | 66.5672   |
-| No log        | 18.0  | 3937 | 0.8907          | 66.6538   |
-| No log        | 19.0  | 4156 | 0.8829          | 66.5278   |
-| No log        | 20.0  | 4375 | 0.9925          | 66.5495   |
-| No log        | 21.0  | 4593 | 0.9656          | 66.5410   |
-| No log        | 22.0  | 4812 | 0.9721          | 66.4741   |
-| No log        | 23.0  | 5031 | 0.9778          | 66.6736   |
-| No log        | 24.0  | 5250 | 1.0032          | 66.5801   |
-| No log        | 25.0  | 5468 | 1.0808          | 66.6122   |
-| No log        | 26.0  | 5687 | 1.0403          | 66.7292   |
-| No log        | 27.0  | 5906 | 1.0388          | 66.5946   |
-| No log        | 28.0  | 6125 | 1.0707          | 66.6240   |
-| No log        | 29.0  | 6343 | 1.0356          | 66.7184   |
-| No log        | 29.9  | 6540 | 1.0664          | 66.6476   |
+| No log        | 0.99  | 54   | 0.5848          | 65.0804   |
+| No log        | 1.97  | 108  | 0.6032          | 65.7419   |
 
 
 ### Framework versions