readme updated

README.md

@@ -25,10 +25,26 @@ More precisely, it was pretrained with the Masked language modeling (MLM) object
 
 This way, the model learns an inner representation of 100 languages that can then be used to extract features useful for downstream tasks: if you have a dataset of labeled sentences for instance, you can train a standard classifier using the features produced by the XLM-RoBERTa model as inputs.
 
-## Training result
+## Training procedure
+
+Fine-tuning was done via the `Trainer` API. Here is the [Colab notebook](https://colab.research.google.com/drive/15LJTckS6gU3RQOmjLqxVNBmbsBdnUEvl?usp=sharing) with the training code.
+
 
-Epoch	  Training_Loss	   Validation_Loss	   Accuracy	    F1
-- 1       0.003000	         0.083116	       0.986136	   0.986332
-- 2	      0.000900	         0.069443	       0.987273	   0.987403
-- 3	      0.087900	         0.067496	       0.988409	   0.988507
+### Training hyperparameters
+
+The following hyperparameters were used during training:
+- learning_rate: 2e-5
+- train_batch_size: 8
+- eval_batch_size: 16
+- optimizer: Adam 
+- evaluation strategy: epoch
+- num_epochs: 3
+- warmup_steps: 100
+
+## Training result
 
+| Training Loss | Epoch | Validation Loss | Accuracy | F1     |
+|:-------------:|:-----:|:---------------:|:--------:|:------:|
+| 0.003000      | 1     |  0.083116       | 0.9861   | 0.9863 |
+| 0.000900      | 2     |  0.069443       | 0.9872   | 0.9874 |
+| 0.087900      | 3     |  0.067496       | 0.9884   | 0.9885 |