Upload folder using huggingface_hub

app/content.py

@@ -61,17 +61,17 @@ displayname2datasetname = {
     'Parliament-Short': 'parliament_short_test',
     'UKUS-News-Short' : 'ukusnews_short_test',
     'Mediacorp-Short' : 'mediacorp_short_test',
-    'YTB-ASR-Batch1(English with Singapore Content)': 'ytb_asr_batch1',
-    'YTB-ASR-Batch2(English with Strong Emotion)': 'ytb_asr_batch2',
-    'YTB-ASR-Batch3-Malay(Malay and Malay-Eng-CodeSwitch)': 'ytb_asr_batch3_ms',
-    'YTB-ASR-Batch3-Malay-Malay-Prompte(Malay and Malay-Eng-CodeSwitch)': 'ytb_asr_batch3_ms_ms_prompt',
+    'YouTube ASR: English with Singapore Content': 'ytb_asr_batch1',
+    'YouTube ASR: English with Strong Emotion': 'ytb_asr_batch2',
+    'YouTube ASR: Malay with English Prompt': 'ytb_asr_batch3_ms',
+    'YouTube ASR: Malay with Malay Prompt': 'ytb_asr_batch3_ms_ms_prompt',
     
-    'SEAME-Dev-Man'   : 'seame_dev_man',
-    'SEAME-Dev-Sge'   : 'seame_dev_sge',
+    'SEAME-Dev-Mandarin'   : 'seame_dev_man',
+    'SEAME-Dev-Singlish'   : 'seame_dev_sge',
 
-    'YTB-SQA-Batch1(English with Singapore Content)': 'ytb_sqa_batch1',
-    'YTB-SDS-Batch1(English with Singapore Content)': 'ytb_sds_batch1',
-    'YTB-PQA-Batch1(English with Singapore Content)': 'ytb_pqa_batch1',
+    'YouTube SQA: English with Singapore Content': 'ytb_sqa_batch1',
+    'YouTube SDS: English with Singapore Content': 'ytb_sds_batch1',
+    'YouTube PQA: English with Singapore Content': 'ytb_pqa_batch1',
 
 }
 
@@ -141,16 +141,17 @@ dataset_diaplay_information = {
     'Parliament-Short': 'Under Development',
     'UKUS-News-Short' : 'Under Development',
     'Mediacorp-Short' : 'Under Development',
-    'YTB-ASR-Batch1(English with Singapore Content)'  : 'YouTube ASR Dataset, English with Singapore Content',
-    'YTB-ASR-Batch2(English with Strong Emotion)'  : 'YouTube ASR Dataset, English with strong emotions',
-    'YTB-ASR-Batch3-Malay(Malay and Malay-Eng-CodeSwitch)': 'YouTube ASR Dataset, Malay and Malay-English CondeSwitch',
-    'YTB-ASR-Batch3-Malay-Malay-Prompte(Malay and Malay-Eng-CodeSwitch)': 'YouTube ASR Dataset, Malay and Malay-English CondeSwitch. Use Malay prompts',
-    'SEAME-Dev-Man'   : 'Under Development',
-    'SEAME-Dev-Sge'   : 'Under Development',
+    'YouTube ASR: English Singapore Content'  : '''\nYouTube Evaluation Dataset for ASR Task: This dataset include English and Singlish with Singapore Content.''',
+    'YouTube ASR: English with Strong Emotion'  : '\nYouTube Evaluation Dataset for ASR Task. English with strong emotions',
+    'YouTube ASR: Malay English Prompt': 'YouTube ASR Dataset, Malay and Malay-English CondeSwitch',
+    'YouTube ASR: Malay with Malay Prompt': 'YouTube ASR Dataset, Malay and Malay-English CondeSwitch. Use Malay prompts',
+    
+    'SEAME-Dev-Mandarin'   : 'Under Development',
+    'SEAME-Dev-Singlish'   : 'Under Development',
 
-    'YTB-SQA-Batch1(English with Singapore Content)': 'Under Development',
-    'YTB-SDS-Batch1(English with Singapore Content)': 'Under Development',
-    'YTB-PQA-Batch1(English with Singapore Content)': 'Under Development',
+    'YouTube SQA: English with Singapore Content': 'Under Development',
+    'YouTube SDS: English with Singapore Content': 'Under Development',
+    'YouTube PQA: English with Singapore Content': 'Under Development',
     
     
                 }

app/draw_diagram.py

@@ -25,7 +25,6 @@ def draw(folder_name, category_name, displayname, metrics, cus_sort=True):
     # Rename to proper display name
     chart_data = chart_data.rename(columns=datasetname2diaplayname)
 
-
     st.markdown("""
                 <style>
                 .stMultiSelect [data-baseweb=select] span {
@@ -52,9 +51,8 @@ def draw(folder_name, category_name, displayname, metrics, cus_sort=True):
     chart_data = chart_data.sort_values(by=[displayname], ascending=cus_sort).dropna(axis=0)
 
     if len(chart_data) == 0: return
-
-
-
+    
+    
     # = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = =
     '''
     Show Table
@@ -222,32 +220,4 @@ def draw(folder_name, category_name, displayname, metrics, cus_sort=True):
             }
 
             value = st_echarts(options=options, events=events, height="500px")
-            
-
-
-
-    # = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = =
-
-    '''
-    Show Examples
-    '''
-
-
-    # Initialize a session state variable for toggling the chart visibility
-    if "show_examples" not in st.session_state:
-        st.session_state.show_examples = False
-
-    # Create a button to toggle visibility
-    if st.button("Show Examples"):
-        st.session_state.show_examples = not st.session_state.show_examples
-
-    if st.session_state.show_examples:
-        
-        st.markdown('To be implemented')
-
-        # # if dataset_name in ['Earnings21-Test', 'Earnings22-Test', 'Tedlium3-Test', 'Tedlium3-Long-form-Test']:
-        # if dataset_name in []:
-        #     pass
-        # else:
-        #     show_examples(category_name, dataset_name, chart_data['Model'].tolist(), display_model_names)
-        
+            

app/pages.py

@@ -2,6 +2,7 @@ import streamlit as st
 from app.draw_diagram import *
 from app.content import *
 from app.summarization import *
+from app.show_examples import *
 
 def dataset_contents(dataset, metrics):
     
@@ -539,16 +540,17 @@ def under_development():
                       'UKUS-News-Short',
                       'Mediacorp-Short',
 
-                      'YTB-ASR-Batch1(English with Singapore Content)',
-                      'YTB-ASR-Batch2(English with Strong Emotion)',
-                      'YTB-ASR-Batch3-Malay(Malay and Malay-Eng-CodeSwitch)',
-                      'YTB-ASR-Batch3-Malay-Malay-Prompte(Malay and Malay-Eng-CodeSwitch)',
-                      'SEAME-Dev-Man',
-                      'SEAME-Dev-Sge',
+                      'YouTube ASR: English Singapore Content',
+                      'YouTube ASR: English with Strong Emotion',
+                      'YouTube ASR: Malay English Prompt',
+                      'YouTube ASR: Malay with Malay Prompt',
+                      
+                      'SEAME-Dev-Mandarin',
+                      'SEAME-Dev-Singlish',
 
-                      'YTB-SQA-Batch1(English with Singapore Content)',
-                      'YTB-SDS-Batch1(English with Singapore Content)',
-                      'YTB-PQA-Batch1(English with Singapore Content)',
+                      'YouTube SQA: English with Singapore Content',
+                      'YouTube SDS: English with Singapore Content',
+                      'YouTube PQA: English with Singapore Content',
 
                       ]
 
@@ -561,6 +563,34 @@ def under_development():
     
     dataset_contents(dataset_diaplay_information[filter_1], 'under_development')
     
+    # = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = =
+
+    '''
+    Show Dataset Examples
+    '''
+
+    # Initialize a session state variable for toggling the chart visibility
+    if "show_dataset_examples" not in st.session_state:
+        st.session_state.show_dataset_examples = False
+
+    # Create a button to toggle visibility
+    if st.button("Show Dataset Examples"):
+        st.session_state.show_dataset_examples = not st.session_state.show_dataset_examples
+
+    if st.session_state.show_dataset_examples:
+        
+        # st.markdown('To be implemented')
+
+        # # if dataset_name in ['Earnings21-Test', 'Earnings22-Test', 'Tedlium3-Test', 'Tedlium3-Long-form-Test']:
+        if filter_1 in []:
+            pass
+        else:
+            try:
+                show_dataset_examples(filter_1)
+            except:
+                st.markdown('To be implemented')
+    # = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = =    
+    
     if filter_1 in [
                       'CNA',
                       'IDPC',
@@ -571,20 +601,22 @@ def under_development():
                       'Parliament-Short',
                       'UKUS-News-Short',
                       'Mediacorp-Short',
-                      'YTB-ASR-Batch1(English with Singapore Content)',
-                      'YTB-ASR-Batch2(English with Strong Emotion)',
-                      'YTB-ASR-Batch3-Malay(Malay and Malay-Eng-CodeSwitch)',
-                      'YTB-ASR-Batch3-Malay-Malay-Prompte(Malay and Malay-Eng-CodeSwitch)',
-                      'SEAME-Dev-Man',
-                      'SEAME-Dev-Sge',
+                      
+                      'YouTube ASR: English Singapore Content',
+                      'YouTube ASR: English with Strong Emotion',
+                      'YouTube ASR: Malay English Prompt',
+                      'YouTube ASR: Malay with Malay Prompt',
+                      
+                      'SEAME-Dev-Mandarin',
+                      'SEAME-Dev-Singlish',
                       ]:
         
         draw('vu', 'under_development_wer', filter_1, 'wer')
 
     elif filter_1 in [
-        'YTB-SQA-Batch1(English with Singapore Content)',
-        'YTB-SDS-Batch1(English with Singapore Content)',
-        'YTB-PQA-Batch1(English with Singapore Content)',
+        'YouTube SQA: English with Singapore Content',
+        'YouTube SDS: English with Singapore Content',
+        'YouTube PQA: English with Singapore Content',
         ]:
         draw('vu', 'under_development_llama3_70b_judge', filter_1, 'llama3_70b_judge')
 

app/show_examples.py

@@ -4,6 +4,79 @@ import numpy as np
 
 import html
 
+from app.content import displayname2datasetname
+
+def show_dataset_examples(display_name):
+    st.divider()
+    dataset_name = displayname2datasetname[display_name]
+    sample_folder = f"./examples/{dataset_name}"
+    
+    # load dataset
+    dataset = datasets.load_from_disk(sample_folder)
+    
+    for index in range(len(dataset)):
+        with st.container():
+            st.markdown(f'##### Example-{index+1}')
+            col1, col2 = st.columns([0.3, 0.7], vertical_alignment="center")
+
+            # with col1:
+            st.audio(f'{sample_folder}/sample_{index}.wav', format="audio/wav")
+                        
+            if dataset_name in ['CN-College-Listen-MCQ-Test', 'DREAM-TTS-MCQ-Test']:
+                
+                choices = dataset[index]['other_attributes']['choices'] 
+                if isinstance(choices, str):
+                    choices_text = choices
+                elif isinstance(choices, list):
+                    choices_text = ' '.join(i for i in choices)
+                
+                question_text = f"""{dataset[index]['instruction']['text']} {choices_text}"""
+            else:
+                question_text = f"""{dataset[index]['instruction']['text']}"""
+            
+            question_text = html.escape(question_text)
+        
+        with st.container():
+            custom_css = """
+                        <style>
+                        .my-container-table, p.my-container-text {
+                        background-color: #fcf8dc;
+                        padding: 10px;
+                        border-radius: 5px;
+                        font-size: 13px;
+                        # height: 50px;
+                        word-wrap: break-word
+                        }
+                        </style>
+                        """
+            st.markdown(custom_css, unsafe_allow_html=True)
+
+            s = f"""<tr>
+                    <td><b>{html.escape(question_text.replace('(A)', '<br>(A)').replace('(B)', '<br>(B)').replace('(C)', '<br>(C)'))}
+                    </td>
+                    <td><b>{html.escape(dataset[index]['answer']['text'])}
+                    </td>
+            </tr>
+            """
+
+            body_details = f"""<table style="table-layout: fixed; width:100%">
+            <thead>
+                <tr style="text-align: center;">
+                    <th style="width:50%">PROMPT</th>
+                    <th style="width:50%">ANSWER</th>
+                </tr>
+            {s}
+            </thead>
+            </table>"""
+            
+            st.markdown(f"""<div class="my-container-table">
+                            {body_details}
+                            </div>""", unsafe_allow_html=True)
+        
+            st.text("")
+    
+    st.divider()
+    
 
 def show_examples(category_name, dataset_name, model_lists, display_model_names):
     st.divider()