Upload folder using huggingface_hub

app/content.py

@@ -0,0 +1,74 @@
+asr_datsets = {'LibriSpeech-Test-Clean': 'aa', 
+                'LibriSpeech-Test-Other': 'bb', 
+                'Common-Voice-15-En-Test': 'cc', 
+                'Peoples-Speech-Test': 'dd', 
+                'GigaSpeech-Test': 'ee', 
+                'Earnings21-Test': 'ff', 
+                'Earnings22-Test': 'gg', 
+                'Tedlium3-Test': 'hh', 
+                'Tedlium3-Longform-Test': 'ii', 
+                'IMDA-Part1-ASR-Test': 'jj', 
+                'IMDA-Part2-ASR-Test': 'kk', 
+                'IMDA-Part3-ASR-Test': 'll', 
+                'IMDA-Part4-ASR-Test': 'mm', 
+                'IMDA-Part5-ASR-Test': 'nn', 
+                'IMDA-Part6-ASR-Test': 'oo'
+                }
+
+sqa_datasets = {'CN-College-Listen-MCQ-Test': 'aa', 
+                'DREAM-TTS-MCQ-Test': 'bb', 
+                'SLUE-P2-SQA5-Test': 'cc', 
+                'Public-SG-Speech-QA-Test': 'dd', 
+                'Spoken-Squad-v1': 'ee'
+                }
+
+si_datasets = {'OpenHermes-Audio-Test': 'aa', 
+               'ALPACA-Audio-Test': 'bb'
+               }
+
+ac_datasets = {
+    'WavCaps-Test': 'aa', 
+    'AudioCaps-Test': 'bb'
+}
+
+asqa_datasets = {
+    'Clotho-AQA-Test': 'aa', 
+    'WavCaps-QA-Test': 'bb', 
+    'AudioCaps-QA-Test': 'cc'
+}
+
+er_datasets = {
+    'IEMOCAP-Emotion-Test': 'aa', 
+    'MELD-Sentiment-Test': 'bb', 
+    'MELD-Emotion-Test': 'cc'
+}
+
+ar_datsets = {
+    'VoxCeleb-Accent-Test': 'aa'
+}
+
+gr_datasets = {
+    'VoxCeleb-Gender-Test': 'aa', 
+    'IEMOCAP-Gender-Test': 'bb'
+}
+
+spt_datasets = {
+    'Covost2-EN-ID-test': 'aa', 
+    'Covost2-EN-ZH-test': 'bb',
+    'Covost2-EN-TA-test': 'cc', 
+    'Covost2-ID-EN-test': 'dd', 
+    'Covost2-ZH-EN-test': 'ee', 
+    'Covost2-TA-EN-test': 'ff'
+}
+
+cnasr_datasets = {
+    'Aishell-ASR-ZH-Test': 'aa'
+}
+
+metrics = {
+    'wer': '11',
+    'llama3_70b_judge_binary': '22',
+    'llama3_70b_judge': '33',
+    'meteor': '44',
+    'bleu': '55'
+}

app/pages.py

@@ -1,5 +1,29 @@
 import streamlit as st
 from app.draw_diagram import *
+from app.content import *
+
+def dataset_contents(dataset, metrics):
+    
+    custom_css = """
+                <style>
+                .my-dataset-info {
+                # background-color: #F9EBEA;
+                # padding: 10px;
+                color: #626567;
+                font-style: italic;
+                font-size: 8px;
+                height: auto;
+                }
+                </style>
+                """
+    st.markdown(custom_css, unsafe_allow_html=True)
+    st.markdown(f"""<div class="my-dataset-info">
+                    <p>DATASET INFORMATION: {dataset}</p>
+                    </div>""", unsafe_allow_html=True)
+    st.markdown(f"""<div class="my-dataset-info">
+                    <p>METRIC INFORMATION: {metrics}</p>
+                    </div>""", unsafe_allow_html=True)
+
 
 def dashboard():
 
@@ -107,9 +131,10 @@ def asr():
     #     sorted = st.selectbox('by', ['Ascending', 'Descending'])
 
     if filter_1:
+        dataset_contents(asr_datsets[filter_1], metrics['wer'])
         draw('su', 'ASR', filter_1, 'wer')
-    else:
-        draw('su', 'ASR', 'LibriSpeech-Test-Clean', 'wer')
+    # else:
+    #     draw('su', 'ASR', 'LibriSpeech-Test-Clean', 'wer')
 
 
     ## examples
@@ -133,11 +158,14 @@ def sqa():
 
     if filter_1:
         if filter_1 in binary:
+            dataset_contents(sqa_datasets[filter_1], metrics['llama3_70b_judge_binary'])
             draw('su', 'SQA', filter_1, 'llama3_70b_judge_binary')
+        
         else:
+            dataset_contents(sqa_datasets[filter_1], metrics['llama3_70b_judge'])
             draw('su', 'SQA', filter_1, 'llama3_70b_judge')
-    else:
-        draw('su', 'SQA', 'CN-College-Listen-Test', 'llama3_70b_judge_binary')
+    # else:
+    #     draw('su', 'SQA', 'CN-College-Listen-Test', 'llama3_70b_judge_binary')
 
 def si():
     st.title("Speech Question Answering")
@@ -151,9 +179,10 @@ def si():
         filter_1 = st.selectbox('Select Dataset', filters_levelone)
 
     if filter_1:
+        dataset_contents(si_datasets[filter_1], metrics['llama3_70b_judge'])
         draw('su', 'SI', filter_1, 'llama3_70b_judge')
-    else:
-        draw('su', 'SI', 'OpenHermes-Audio-Test', 'llama3_70b_judge')
+    # else:
+    #     draw('su', 'SI', 'OpenHermes-Audio-Test', 'llama3_70b_judge')
 
 def ac():
     st.title("Audio Captioning")
@@ -181,9 +210,10 @@ def ac():
     #     sorted = st.selectbox('by', ['Ascending', 'Descending'])
 
     if filter_1 or metric:
+        dataset_contents(ac_datasets[filter_1], metrics[metric.lower().replace('-', '_')])
         draw('asu', 'AC',filter_1, metric.lower().replace('-', '_'))
-    else:
-        draw('asu', 'AC', 'WavCaps-Test', 'llama3_70b_judge')
+    # else:
+    #     draw('asu', 'AC', 'WavCaps-Test', 'llama3_70b_judge')
 
 def asqa():
     st.title("Audio Scene Question Answering")
@@ -198,9 +228,10 @@ def asqa():
         filter_1 = st.selectbox('Select Dataset', filters_levelone)
     
     if filter_1:
+        dataset_contents(asqa_datasets[filter_1], metrics['llama3_70b_judge'])
         draw('asu', 'AQA',filter_1, 'llama3_70b_judge')
-    else:
-        draw('asu', 'AQA', 'Clotho-AQA-Test', 'llama3_70b_judge')
+    # else:
+    #     draw('asu', 'AQA', 'Clotho-AQA-Test', 'llama3_70b_judge')
 
 def er():
     st.title("Emotion Recognition")
@@ -208,7 +239,7 @@ def er():
     filters_levelone = ['IEMOCAP-Emotion-Test', 
                         'MELD-Sentiment-Test', 
                         'MELD-Emotion-Test']
-    sort_leveltwo = []
+    # sort_leveltwo = []
     
     left, center, _, middle,right = st.columns([0.2, 0.2, 0.2, 0.2 ,0.2])
     
@@ -231,9 +262,10 @@ def er():
     #     sorted = st.selectbox('by', ['Ascending', 'Descending'])
 
     if filter_1:
+        dataset_contents(er_datasets[filter_1], metrics['llama3_70b_judge_binary'])
         draw('vu', 'ER', filter_1, 'llama3_70b_judge_binary')
-    else:
-        draw('vu', 'ER', 'IEMOCAP-Emotion-Test', 'llama3_70b_judge_binary')
+    # else:
+    #     draw('vu', 'ER', 'IEMOCAP-Emotion-Test', 'llama3_70b_judge_binary')
 
 def ar():
     st.title("Accent Recognition")
@@ -247,9 +279,9 @@ def ar():
 
 
     if filter_1:
+        dataset_contents(ar_datsets[filter_1], metrics['llama3_70b_judge'])
         draw('vu', 'AR', filter_1, 'llama3_70b_judge')
-    else:
-        draw('vu', 'AR', 'VoxCeleb-Accent-Test', 'llama3_70b_judge')
+
 
 def gr():
     st.title("Emotion Recognition")
@@ -263,9 +295,10 @@ def gr():
         filter_1 = st.selectbox('Select Dataset', filters_levelone)
     
     if filter_1:
+        dataset_contents(gr_datasets[filter_1], metrics['llama3_70b_judge_binary'])
         draw('vu', 'GR', filter_1, 'llama3_70b_judge_binary')
-    else:
-        draw('vu', 'GR', 'VoxCeleb1-Gender-Test', 'llama3_70b_judge_binary')
+    # else:
+    #     draw('vu', 'GR', 'VoxCeleb1-Gender-Test', 'llama3_70b_judge_binary')
 
 def spt():
     st.title("Speech Translation")
@@ -283,9 +316,10 @@ def spt():
         filter_1 = st.selectbox('Select Dataset', filters_levelone)
     
     if filter_1:
+        dataset_contents(spt_datasets[filter_1], metrics['bleu'])
         draw('su', 'ST', filter_1, 'bleu')
-    else:
-        draw('su', 'ST', 'Covost2-EN-ID-test', 'bleu')
+    # else:
+    #     draw('su', 'ST', 'Covost2-EN-ID-test', 'bleu')
 
 def cnasr():
     st.title("Chinese Automatic Speech Recognition")
@@ -298,6 +332,7 @@ def cnasr():
         filter_1 = st.selectbox('Select Dataset', filters_levelone)
     
     if filter_1:
+        dataset_contents(cnasr_datasets[filter_1], metrics['wer'])
         draw('su', 'CNASR', filter_1, 'wer')
-    else:
-        draw('su', 'CNASR', 'Aishell-ASR-ZH-Test', 'wer')
+    # else:
+    #     draw('su', 'CNASR', 'Aishell-ASR-ZH-Test', 'wer')