update model card README.md

README.md

@@ -18,8 +18,8 @@ should probably proofread and complete it, then remove this comment. -->
 
 This model is a fine-tuned version of [roberta-base](https://huggingface.co/roberta-base) on the glue dataset.
 It achieves the following results on the evaluation set:
-- Loss: 0.5684
-- Matthews Correlation: 0.6113
+- Loss: 0.5127
+- Matthews Correlation: 0.5815
 
 ## Model description
 
@@ -44,8 +44,8 @@ The following hyperparameters were used during training:
 - seed: 42
 - distributed_type: IPU
 - gradient_accumulation_steps: 16
-- total_train_batch_size: 16
-- total_eval_batch_size: 5
+- total_train_batch_size: 64
+- total_eval_batch_size: 20
 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
 - lr_scheduler_type: linear
 - num_epochs: 5
@@ -55,11 +55,11 @@ The following hyperparameters were used during training:
 
 | Training Loss | Epoch | Step | Validation Loss | Matthews Correlation |
 |:-------------:|:-----:|:----:|:---------------:|:--------------------:|
-| 0.4472        | 1.0   | 534  | 0.3931          | 0.5853               |
-| 0.3492        | 2.0   | 1068 | 0.4504          | 0.5900               |
-| 0.0294        | 3.0   | 1602 | 0.4812          | 0.5866               |
-| 0.1438        | 4.0   | 2136 | 0.4937          | 0.5950               |
-| 0.0249        | 5.0   | 2670 | 0.5684          | 0.6113               |
+| 0.3138        | 1.0   | 133  | 0.4104          | 0.5538               |
+| 0.2082        | 2.0   | 266  | 0.4849          | 0.5402               |
+| 0.1687        | 3.0   | 399  | 0.5127          | 0.5815               |
+| 0.0865        | 4.0   | 532  | 0.5752          | 0.5661               |
+| 0.1008        | 5.0   | 665  | 0.5952          | 0.5764               |
 
 
 ### Framework versions