Update README.md

README.md

@@ -14,7 +14,17 @@ library_name: transformers
 
 This model is based on our pretrained [5CD-AI/visobert-14gb-corpus](https://huggingface.co/5CD-AI/visobert-14gb-corpus), which has been continuously trained on a 14GB dataset of Vietnamese social content.
 
-Our model is fine-tuned on <b>120K</b> Vietnamese sentiment datasets, including comments and reviews from e-commerce platforms, social media, and forums.
+Our model is fine-tuned on <b>120K</b> Vietnamese sentiment datasets, including comments and reviews from e-commerce platforms, social media, and forums
+
+Our model get over performace in datasets: 
+- SA-VLSP2016
+- AIVIVN-2019
+- UIT-VSFC
+- UIT-VSMEC
+- UIT-ViCTSD
+- UIT-ViOCD
+- UIT-ViSFD
+- Vi-amazon-polar
 
 The model will give softmax outputs for three labels.
 
@@ -27,7 +37,7 @@ The model will give softmax outputs for three labels.
 ```
 
 ## Dataset
-
+Our training dataset. With UIT-VSMEC, UIT-ViCTSD, VOZ-HSD, we re-label the dataset with Gemini 1.5 Flash API follow the 3 labels. 
 <table border="2">
     <tr align="center">
         <th rowspan="2">Dataset</th>
@@ -83,7 +93,7 @@ The model will give softmax outputs for three labels.
         <td>-</td>
     </tr>
       <tr align="center">
-        <td align="left">UIT-VSMEC</td>
+        <td align="left">UIT-VSMEC (Gemini-label)</td>
         <td>3219</td>
         <td>1665</td>
         <td>594</td>
@@ -108,7 +118,7 @@ The model will give softmax outputs for three labels.
         <td>-</td>
     </tr>
       <tr align="center">
-        <td align="left">UIT-ViCTSD</td>
+        <td align="left">UIT-ViCTSD (Gemini-label)</td>
         <td>3370</td>
         <td>2615</td>
         <td>933</td>
@@ -156,7 +166,7 @@ The model will give softmax outputs for three labels.
         <td>-</td>
     </tr>
       <tr align="center">
-        <td align="left">Ecommerce-reviews</td>
+        <td align="left">Tiki-reviews</td>
         <td>20093</td>
         <td>6669</td>
         <td>4698</td>
@@ -168,16 +178,16 @@ The model will give softmax outputs for three labels.
         <td>-</td>
     </tr>
       <tr align="center">
-        <td align="left">VOZ-HSD</td>
+        <td align="left">VOZ-HSD (Gemini-label)</td>
         <td>2676</td>
         <td>1213</td>
         <td>1071</td>
-        <td>1068</td>
-        <td>495</td>
-        <td>420</td>
-        <td>511</td>
-        <td>224</td>
-        <td>199</td>
+        <td>-</td>
+        <td>-</td>
+        <td>-</td>
+        <td>-</td>
+        <td>-</td>
+        <td>-</td>
     </tr>
       <tr align="center">
         <td align="left">Vietnamese-amazon-polarity</td>
@@ -200,8 +210,8 @@ The model will give softmax outputs for three labels.
             <td colspan=4><b>SA-VLSP2016</td>
             <td colspan=4><b>AIVIVN-2019</td>
             <td colspan=4><b>UIT-VSFC</td>
-            <td colspan=4><b>UIT-VSMEC</td>
-            <td colspan=4><b>UIT-ViCTSD</td>
+            <td colspan=4><b>UIT-VSMEC (Gemini-label)</td>
+            <td colspan=4><b>UIT-ViCTSD (Gemini-label)</td>
         </tr>
         <tr align="center">
             <td><b>Acc</td>
@@ -281,7 +291,6 @@ The model will give softmax outputs for three labels.
             <td rowspan=2><b>Model</td>
             <td colspan=4><b>UIT-ViOCD</td>
             <td colspan=4><b>UIT-ViSFD</td>
-            <td colspan=4><b>VOZ-HSD</td>
             <td colspan=4><b>Vi-amazon-polar</td>
         </tr>
         <tr align="center">
@@ -297,14 +306,11 @@ The model will give softmax outputs for three labels.
             <td><b>Prec</td>
             <td><b>Recall</td>
             <td><b>WF1</td>
-            <td><b>Acc</td>
-            <td><b>Prec</td>
-            <td><b>Recall</td>
-            <td><b>WF1</td>
         </tr>
         <tr align="center">
         <tr align="center">
             <td align="left">wonrax/phobert-base-vietnamese-sentiment</td>
+            <td>74.68</td>
             <td>87.14</td>
             <td>74.68</td>
             <td>78.13</td>
@@ -312,10 +318,6 @@ The model will give softmax outputs for three labels.
             <td>67.95</td>
             <td>67.90</td>
             <td>66.98</td>
-            <td>51.89</td>
-            <td>60.18</td>
-            <td>51.89</td>
-            <td>53.61</td>
             <td>61.40</td>
             <td>76.53</td>
             <td>61.40</td>
@@ -331,10 +333,6 @@ The model will give softmax outputs for three labels.
             <td><b>93.20</td>
             <td><b>93.26</td>
             <td><b>93.21</td>
-            <td><b>67.78</td>
-            <td><b>69.82</td>
-            <td><b>67.78</td>
-            <td><b>68.39</td>
             <td><b>89.90</td>
             <td><b>90.13</td>
             <td><b>89.90</td>