File size: 8,151 Bytes
f5625dd
 
 
5e2340a
 
f5625dd
62da12f
 
f7a3f90
 
 
 
 
 
 
 
 
 
 
 
 
 
 
5e2340a
 
 
 
 
 
 
 
 
f7a3f90
 
 
 
 
 
 
 
 
 
 
 
 
 
5e2340a
 
f7a3f90
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
16943d7
 
 
 
 
 
f7a3f90
 
 
 
 
 
 
 
 
5e2340a
f7a3f90
62da12f
8764f41
f5625dd
 
308cc2e
f5625dd
 
 
a30610b
8764f41
f5625dd
 
4237375
 
a30610b
4237375
 
 
 
 
 
 
 
 
 
 
62da12f
 
f5625dd
a30610b
f5625dd
 
 
8764f41
 
f5625dd
 
8764f41
f5625dd
4237375
f5625dd
 
 
 
 
 
4237375
 
8764f41
62da12f
8764f41
62da12f
8764f41
 
 
f5625dd
 
 
62da12f
 
 
 
f5625dd
8764f41
 
 
 
 
62da12f
f5625dd
 
 
 
 
 
62da12f
 
 
f5625dd
62da12f
 
 
 
f5625dd
8764f41
62da12f
 
f5625dd
 
 
 
8764f41
f5625dd
4237375
f5625dd
 
 
62da12f
 
f5625dd
 
 
 
 
 
 
 
 
 
a30610b
 
f5625dd
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
import streamlit as st
import datasets
import numpy as np
import io
import soundfile as sf

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:
                # Convert the NumPy array to a WAV file in memory
                bytes_io = io.BytesIO()
                sf.write(bytes_io, dataset[index]['context']['audio']['array'], dataset[index]['context']['audio']['sampling_rate'], format='WAV')
                bytes_io.seek(0)
                # Play audio in Streamlit
                st.audio(bytes_io, format='audio/wav')
                # 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():
        with col2:
            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)

            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>
                <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>
            </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()
    sample_folder = f"./examples/{category_name}/{dataset_name}"
    
    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)
            
            # st.divider()
            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)

                model_lists.sort()

                s = f"""<tr>
                       <td><b>REFERENCE</td>
                       <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>
                """
                if dataset_name in ['CN-College-Listen-MCQ-Test', 'DREAM-TTS-MCQ-Test']:
                    for model in model_lists:
                        try:

                            model_prediction = dataset[index][model]['model_prediction']
                            model_prediction = model_prediction.replace('<','').replace('>','').replace('\n','(newline)').replace('*','')

                            s += f"""<tr>
                                <td>{display_model_names[model]}</td>
                                <td>
                                    {dataset[index][model]['text'].replace('Choices:', '<br>Choices:').replace('(A)', '<br>(A)').replace('(B)', '<br>(B)').replace('(C)', '<br>(C)') 
                                     }
                                    </td>
                                <td>{html.escape(model_prediction)}</td>
                            </tr>"""
                        except:
                            print(f"{model} is not in {dataset_name}")
                            continue
                else:
                    for model in model_lists:

                        print(dataset[index][model]['model_prediction'])

                        try:

                            model_prediction = dataset[index][model]['model_prediction']
                            model_prediction = model_prediction.replace('<','').replace('>','').replace('\n','(newline)').replace('*','')

                            s += f"""<tr>
                                <td>{display_model_names[model]}</td>
                                <td>{html.escape(dataset[index][model]['text'])}</td>
                                <td>{html.escape(model_prediction)}</td>
                            </tr>"""
                        except:
                            print(f"{model} is not in {dataset_name}")
                            continue

                
                body_details = f"""<table style="table-layout: fixed; width:100%">
                <thead>
                    <tr style="text-align: center;">
                        <th style="width:20%">MODEL</th>
                        <th style="width:30%">QUESTION</th>
                        <th style="width:50%">MODEL PREDICTION</th>
                    </tr>
                {s}
                </thead>
                </table>"""
                
                st.markdown(f"""<div class="my-container-table">
                                {body_details}
                                </div>""", unsafe_allow_html=True)
            
                st.text("")
        
        st.divider()