update model card README.md

README.md

@@ -19,12 +19,12 @@ should probably proofread and complete it, then remove this comment. -->
 
 This model is a fine-tuned version of [digitalepidemiologylab/covid-twitter-bert-v2](https://huggingface.co/digitalepidemiologylab/covid-twitter-bert-v2) on an unknown dataset.
 It achieves the following results on the evaluation set:
-- Loss: 0.3048
-- Accuracy: 0.884
-- Precision: 0.8043
-- Recall: 0.6491
-- F1-weighted: 0.8794
-- F1: 0.7184
+- Loss: 0.3348
+- Accuracy: 0.888
+- Precision: 0.7843
+- Recall: 0.7018
+- F1-weighted: 0.8857
+- F1: 0.7407
 
 ## Model description
 
@@ -47,19 +47,20 @@ The following hyperparameters were used during training:
 - train_batch_size: 16
 - eval_batch_size: 2
 - seed: 42
-- gradient_accumulation_steps: 6
-- total_train_batch_size: 96
+- gradient_accumulation_steps: 8
+- total_train_batch_size: 128
 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
 - lr_scheduler_type: linear
 - lr_scheduler_warmup_ratio: 0.05
-- num_epochs: 6
+- num_epochs: 12
 
 ### Training results
 
 | Training Loss | Epoch | Step | Validation Loss | Accuracy | Precision | Recall | F1-weighted | F1     |
 |:-------------:|:-----:|:----:|:---------------:|:--------:|:---------:|:------:|:-----------:|:------:|
-| 0.4786        | 2.73  | 50   | 0.3590          | 0.864    | 0.7018    | 0.7018 | 0.864       | 0.7018 |
-| 0.265         | 5.45  | 100  | 0.3048          | 0.884    | 0.8043    | 0.6491 | 0.8794      | 0.7184 |
+| 0.4411        | 3.64  | 50   | 0.3396          | 0.876    | 0.8611    | 0.5439 | 0.8652      | 0.6667 |
+| 0.1872        | 7.27  | 100  | 0.3182          | 0.876    | 0.6912    | 0.8246 | 0.8796      | 0.7520 |
+| 0.0724        | 10.91 | 150  | 0.3348          | 0.888    | 0.7843    | 0.7018 | 0.8857      | 0.7407 |
 
 
 ### Framework versions