Update README.md

README.md

@@ -13,27 +13,27 @@ widget:
 A fine-tuned model for fact-classification task on English tweets about COVID-19/vaccine.
 
 ## Intended uses & limitations
+You can classify if the input tweet (or any others statement) about COVID-19/vaccine is true, false or misleading.
+Note that since this model was trained with data up to May 2020, the most recent information may not be reflected.
 
 #### How to use
-Load pipeline
+You can use this model directly on this page or using `transformers` in python.
+
+- Load pipeline and implement with input sequence
 ```python
 from transformers import pipeline
 pipe = pipeline("sentiment-analysis", model = "ans/vaccinating-covid-tweets")
-seq = 'The Pfizer-BioNTech COVID-19 vaccine is authorized for use under an Emergency Use Authorization (EUA) for active immunization to prevent COVID-19 caused by SARS-CoV-2 in individuals 16 years of age and older'
-
+seq = "COVID-19 vaccines are safe and effective."
 pipe(seq)
 ```
 
-Expected output
-```
-seq = "There is no FDA-approved vaccine to prevent COVID-19."
-pipe(seq)
+- Expected output
+```python
+[{'label': 'true', 'score': 0.9987803101539612}]
 ```
-[{'label': 'true', 'score': 0.9996959567070007}]
 
 #### Limitations and bias
-
-Provide examples of latent issues and potential remediations.
+To conservatively classify whether an input sequence is true or not, the model may have predictions biased toward false/misleading.
 
 ## Training data & Procedure
 
@@ -62,7 +62,10 @@ Provide examples of latent issues and potential remediations.
   - Misleading: misleading, exaggerated, out of context, needs context
   - True: true, correct
 
-## Eval results
+## Evaluation results
+| Training loss | Validation loss | Training accuracy | Validation accuracy |
+| --- | --- | --- | --- |
+| 0.1062 | 0.1006 | 96.3% | 94.5% |
 
 # Contributors
 - This page is a part of final team project from MLDL for DS class at SNU