Upload folder using huggingface_hub

app/content.py

@@ -1,143 +1,156 @@
 
-dataname_column_rename_in_table = {
-    'librispeech_test_clean'       : 'LibriSpeech-Clean',
-    'librispeech_test_other'       : 'LibriSpeech-Other',
-    'common_voice_15_en_test'      : 'CommonVoice-15-EN',
-    'peoples_speech_test'          : 'Peoples-Speech',
-    'gigaspeech_test'              : 'GigaSpeech-1',
-    'earnings21_test'              : 'Earnings-21',
-    'earnings22_test'              : 'Earnings-22',
-    'tedlium3_test'                : 'TED-LIUM-3',
-    'tedlium3_long_form_test'      : 'TED-LIUM-3-Long',
-    'aishell_asr_zh_test'          : 'Aishell-ASR-ZH',
-    'covost2_en_id_test'           : 'CoVoST2-EN-ID',
-    'covost2_en_zh_test'           : 'CoVoST2-EN-ZH',
-    'covost2_en_ta_test'           : 'CoVoST2-EN-TA',
-    'covost2_id_en_test'           : 'CoVoST2-ID-EN',
-    'covost2_zh_en_test'           : 'CoVoST2-ZH-EN',
-    'covost2_ta_en_test'           : 'CoVoST2-TA-EN',
-    'cn_college_listen_mcq_test'   : 'CN-College-Listen-MCQ',
-    'dream_tts_mcq_test'           : 'DREAM-TTS-MCQ',
-    'slue_p2_sqa5_test'            : 'SLUE-P2-SQA5',
-    'public_sg_speech_qa_test'     : 'Public-SG-Speech-QA',
-    'spoken_squad_test'            : 'Spoken-SQuAD',
-    'openhermes_audio_test'        : 'OpenHermes-Audio',
-    'alpaca_audio_test'            : 'ALPACA-Audio',
-    'wavcaps_test'                 : 'WavCaps',
-    'audiocaps_test'               : 'AudioCaps',
-    'clotho_aqa_test'              : 'Clotho-AQA',
-    'wavcaps_qa_test'              : 'WavCaps-QA',
-    'audiocaps_qa_test'            : 'AudioCaps-QA',
-    'voxceleb_accent_test'         : 'VoxCeleb-Accent',
-    'voxceleb_gender_test'         : 'VoxCeleb-Gender',
-    'iemocap_gender_test'          : 'IEMOCAP-Gender',
-    'iemocap_emotion_test'         : 'IEMOCAP-Emotion',
-    'meld_sentiment_test'          : 'MELD-Sentiment',
-    'meld_emotion_test'            : 'MELD-Emotion',
-    'imda_part1_asr_test'          : 'IMDA-Part1-ASR',
-    'imda_part2_asr_test'          : 'IMDA-Part2-ASR',
-    'imda_part3_30s_asr_test'      : 'IMDA-Part3-30s-ASR',
-    'imda_part4_30s_asr_test'      : 'IMDA-Part4-30s-ASR',
-    'imda_part5_30s_asr_test'      : 'IMDA-Part5-30s-ASR',
-    'imda_part6_30s_asr_test'      : 'IMDA-Part6-30s-ASR',
-    'muchomusic_test'              : 'MuChoMusic',
-    'imda_part3_30s_sqa_human_test': 'MNSC-PART3-SQA',
-    'imda_part4_30s_sqa_human_test': 'MNSC-PART4-SQA',
-    'imda_part5_30s_sqa_human_test': 'MNSC-PART5-SQA',
-    'imda_part6_30s_sqa_human_test': 'MNSC-PART6-SQA',
-
-   
-}
-
-asr_datsets = {'LibriSpeech-Test-Clean': 'A clean, high-quality testset of the LibriSpeech dataset, used for ASR testing.', 
-                'LibriSpeech-Test-Other' : 'A more challenging, noisier testset of the LibriSpeech dataset for ASR testing.',
-                'Common-Voice-15-En-Test': 'Test set from the Common Voice project, which is a crowd-sourced, multilingual speech dataset.',
-                'Peoples-Speech-Test'    : 'A large-scale, open-source speech recognition dataset, with diverse accents and domains.',
-                'GigaSpeech-Test'        : 'A large-scale ASR dataset with diverse audio sources like podcasts, interviews, etc.',
-                'Earnings21-Test'        : 'ASR test dataset focused on earnings calls from 2021, with professional speech and financial jargon.',
-                'Earnings22-Test'        : 'Similar to Earnings21, but covering earnings calls from 2022.',
-                'Tedlium3-Test'          : 'A test set derived from TED talks, covering diverse speakers and topics.',
-                'Tedlium3-Long-form-Test': 'A longer version of the TED-LIUM dataset, containing extended audio samples. This poses challenges to existing fusion methods in handling long audios. However, it provides benchmark for future development.',
-                }
-
-singlish_asr_datasets = {
-                'IMDA-Part1-ASR-Test'    : 'Speech recognition test data from the IMDA NSC project, Part 1.',
-                'IMDA-Part2-ASR-Test'    : 'Speech recognition test data from the IMDA NSC project, Part 2.',
-                'IMDA-Part3-30s-ASR-Test': 'Speech recognition test data from the IMDA NSC project, Part 3.',
-                'IMDA-Part4-30s-ASR-Test': 'Speech recognition test data from the IMDA NSC project, Part 4.',
-                'IMDA-Part5-30s-ASR-Test': 'Speech recognition test data from the IMDA NSC project, Part 5.',
-                'IMDA-Part6-30s-ASR-Test': 'Speech recognition test data from the IMDA NSC project, Part 6.'
-                }
-
-sqa_datasets = {'CN-College-Listen-MCQ-Test': 'Chinese College English Listening Test, with multiple-choice questions.', 
-                'DREAM-TTS-MCQ-Test'      : 'DREAM dataset for spoken question-answering, derived from textual data and synthesized speech.',
-                'SLUE-P2-SQA5-Test'       : 'Spoken Language Understanding Evaluation (SLUE) dataset, part 2, focused on QA tasks.',
-                'Public-SG-Speech-QA-Test': 'Public dataset for speech-based question answering, gathered from Singapore.',
-                'Spoken-Squad-Test'       : 'Spoken SQuAD dataset, based on the textual SQuAD dataset, converted into audio.'
-                }
-
-sqa_singlish_datasets = {
-                'MNSC-PART3-SQA': 'Multitak National Speech Corpus (MNSC) dataset, Question answering task, Part 3.',
-                'MNSC-PART4-SQA': 'Multitak National Speech Corpus (MNSC) dataset, Question answering task, Part 4.',
-                'MNSC-PART5-SQA': 'Multitak National Speech Corpus (MNSC) dataset, Question answering task, Part 5.',
-                'MNSC-PART6-SQA': 'Multitak National Speech Corpus (MNSC) dataset, Question answering task, Part 6.',
-                }
-
-si_datasets = {
-               'OpenHermes-Audio-Test': 'Test set for spoken instructions. Synthesized from the OpenHermes dataset.',
-               'ALPACA-Audio-Test'    : 'Spoken version of the ALPACA dataset, used for evaluating instruction following in audio.'
-               }
-
-ac_datasets = {
-    'WavCaps-Test'  : 'WavCaps is a dataset for testing audio captioning, where models generate textual descriptions of audio clips.',
-    'AudioCaps-Test': 'AudioCaps dataset, used for generating captions from general audio events.'
-}
-
-asqa_datasets = {
-    'Clotho-AQA-Test'  : 'Clotho dataset adapted for audio-based question answering, containing audio clips and questions.',
-    'WavCaps-QA-Test'  : 'Question-answering test dataset derived from WavCaps, focusing on audio content.',
-    'AudioCaps-QA-Test': 'AudioCaps adapted for question-answering tasks, using audio events as input for Q&A.'
-}
+displayname2datasetname = {
+    'LibriSpeech-Clean'    : 'librispeech_test_clean',
+    'LibriSpeech-Other'    : 'librispeech_test_other',
+    'CommonVoice-15-EN'    : 'common_voice_15_en_test',
+    'Peoples-Speech'       : 'peoples_speech_test',
+    'GigaSpeech-1'         : 'gigaspeech_test',
+    'Earnings-21'          : 'earnings21_test',
+    'Earnings-22'          : 'earnings22_test',
+    'TED-LIUM-3'           : 'tedlium3_test',
+    'TED-LIUM-3-LongForm'  : 'tedlium3_long_form_test',
+    'AISHELL-ASR-ZH'       : 'aishell_asr_zh_test',
+    'CoVoST2-EN-ID'        : 'covost2_en_id_test',
+    'CoVoST2-EN-ZH'        : 'covost2_en_zh_test',
+    'CoVoST2-EN-TA'        : 'covost2_en_ta_test',
+    'CoVoST2-ID-EN'        : 'covost2_id_en_test',
+    'CoVoST2-ZH-EN'        : 'covost2_zh_en_test',
+    'CoVoST2-TA-EN'        : 'covost2_ta_en_test',
+    'CN-College-Listen-MCQ': 'cn_college_listen_mcq_test',
+    'DREAM-TTS-MCQ'        : 'dream_tts_mcq_test',
+    'SLUE-P2-SQA5'         : 'slue_p2_sqa5_test',
+    'Public-SG-Speech-QA'  : 'public_sg_speech_qa_test',
+    'Spoken-SQuAD'         : 'spoken_squad_test',
+    'OpenHermes-Audio'     : 'openhermes_audio_test',
+    'ALPACA-Audio'         : 'alpaca_audio_test',
+    'WavCaps'              : 'wavcaps_test',
+    'AudioCaps'            : 'audiocaps_test',
+    'Clotho-AQA'           : 'clotho_aqa_test',
+    'WavCaps-QA'           : 'wavcaps_qa_test',
+    'AudioCaps-QA'         : 'audiocaps_qa_test',
+    'VoxCeleb-Accent'      : 'voxceleb_accent_test',
+    'MNSC-AR-Sentence'     : 'imda_ar_sentence',
+    'MNSC-AR-Dialogue'     : 'imda_ar_dialogue',
+    'VoxCeleb-Gender'      : 'voxceleb_gender_test',
+    'IEMOCAP-Gender'       : 'iemocap_gender_test',
+    'IEMOCAP-Emotion'      : 'iemocap_emotion_test',
+    'MELD-Sentiment'       : 'meld_sentiment_test',
+    'MELD-Emotion'         : 'meld_emotion_test',
+    'MuChoMusic'           : 'muchomusic_test',
+    'MNSC-PART1-ASR'       : 'imda_part1_asr_test',
+    'MNSC-PART2-ASR'       : 'imda_part2_asr_test',
+    'MNSC-PART3-ASR'       : 'imda_part3_30s_asr_test',
+    'MNSC-PART4-ASR'       : 'imda_part4_30s_asr_test',
+    'MNSC-PART5-ASR'       : 'imda_part5_30s_asr_test',
+    'MNSC-PART6-ASR'       : 'imda_part6_30s_asr_test',
+    'MNSC-PART3-SQA'       : 'imda_part3_30s_sqa_human_test',
+    'MNSC-PART4-SQA'       : 'imda_part4_30s_sqa_human_test',
+    'MNSC-PART5-SQA'       : 'imda_part5_30s_sqa_human_test',
+    'MNSC-PART6-SQA'       : 'imda_part6_30s_sqa_human_test',
+    'MNSC-PART3-SDS'       : 'imda_part3_30s_ds_human_test',
+    'MNSC-PART4-SDS'       : 'imda_part4_30s_ds_human_test',
+    'MNSC-PART5-SDS'       : 'imda_part5_30s_ds_human_test',
+    'MNSC-PART6-SDS'       : 'imda_part6_30s_ds_human_test',
+
+    'CNA'             : 'cna_test',
+    'IDPC'            : 'idpc_test',
+    'Parliament'      : 'parliament_test',
+    'UKUS-News'       : 'ukusnews_test',
+    'Mediacorp'       : 'mediacorp_test',
+    'IDPC-Short'      : 'idpc_short_test',
+    'Parliament-Short': 'parliament_short_test',
+    'UKUS-News-Short' : 'ukusnews_short_test',
+    'Mediacorp-Short' : 'mediacorp_short_test',
+    'YTB-ASR-Batch1'  : 'ytb_asr_batch1',
+    'YTB-ASR-Batch2'  : 'ytb_asr_batch2',
+    'SEAME-Dev-Man'   : 'seame_dev_man',
+    'SEAME-Dev-Sge'   : 'seame_dev_sge',
+
+    'YTB-SQA-Batch1': 'ytb_sqa_batch1',
+    'YTB-SDS-Batch1': 'ytb_sds_batch1',
+    'YTB-PQA-Batch1': 'ytb_pqa_batch1',
 
-er_datasets = {
-    'IEMOCAP-Emotion-Test': 'Emotion recognition test data from the IEMOCAP dataset, focusing on identifying emotions in speech.',
-    'MELD-Sentiment-Test' : 'Sentiment recognition from speech using the MELD dataset, classifying positive, negative, or neutral sentiments.',
-    'MELD-Emotion-Test'   : 'Emotion classification in speech using MELD, detecting specific emotions like happiness, anger, etc.'
 }
 
-ar_datsets = {
-    'VoxCeleb-Accent-Test': 'Test dataset for accent recognition, based on VoxCeleb, a large speaker identification dataset.'
-}
-
-gr_datasets = {
-    'VoxCeleb-Gender-Test': 'Test dataset for gender classification, also derived from VoxCeleb.',
-    'IEMOCAP-Gender-Test' : 'Gender classification based on the IEMOCAP dataset.'
-}
+datasetname2diaplayname = {datasetname: displayname for displayname, datasetname in displayname2datasetname.items()}
+
+
+dataset_diaplay_information = {
+    'LibriSpeech-Clean'    : 'A clean, high-quality testset of the LibriSpeech dataset, used for ASR testing.',
+    'LibriSpeech-Other'    : 'A more challenging, noisier testset of the LibriSpeech dataset for ASR testing.',
+    'CommonVoice-15-EN'    : 'Test set from the Common Voice project, which is a crowd-sourced, multilingual speech dataset.',
+    'Peoples-Speech'       : 'A large-scale, open-source speech recognition dataset, with diverse accents and domains.',
+    'GigaSpeech-1'         : 'A large-scale ASR dataset with diverse audio sources like podcasts, interviews, etc.',
+    'Earnings-21'          : 'ASR test dataset focused on earnings calls from 2021, with professional speech and financial jargon.',
+    'Earnings-22'          : 'Similar to Earnings21, but covering earnings calls from 2022.',
+    'TED-LIUM-3'           : 'A test set derived from TED talks, covering diverse speakers and topics.',
+    'TED-LIUM-3-LongForm'  : 'A longer version of the TED-LIUM dataset, containing extended audio samples. This poses challenges to existing fusion methods in handling long audios. However, it provides benchmark for future development.',
+    'AISHELL-ASR-ZH'       : 'ASR test dataset for Mandarin Chinese, based on the Aishell dataset.',
+    'CoVoST2-EN-ID'        : 'CoVoST 2 dataset for speech translation from English to Indonesian.',
+    'CoVoST2-EN-ZH'        : 'CoVoST 2 dataset for speech translation from English to Chinese.',
+    'CoVoST2-EN-TA'        : 'CoVoST 2 dataset for speech translation from English to Tamil.',
+    'CoVoST2-ID-EN'        : 'CoVoST 2 dataset for speech translation from Indonesian to English.',
+    'CoVoST2-ZH-EN'        : 'CoVoST 2 dataset for speech translation from Chinese to English.',
+    'CoVoST2-TA-EN'        : 'CoVoST 2 dataset for speech translation from Tamil to English.',
+    'CN-College-Listen-MCQ': 'Chinese College English Listening Test, with multiple-choice questions.',
+    'DREAM-TTS-MCQ'        : 'DREAM dataset for spoken question-answering, derived from textual data and synthesized speech.',
+    'SLUE-P2-SQA5'         : 'Spoken Language Understanding Evaluation (SLUE) dataset, part 2, focused on QA tasks.',
+    'Public-SG-Speech-QA'  : 'Public dataset for speech-based question answering, gathered from Singapore.',
+    'Spoken-SQuAD'         : 'Spoken SQuAD dataset, based on the textual SQuAD dataset, converted into audio.',
+    'OpenHermes-Audio'     : 'Test set for spoken instructions. Synthesized from the OpenHermes dataset.',
+    'ALPACA-Audio'         : 'Spoken version of the ALPACA dataset, used for evaluating instruction following in audio.',
+    'WavCaps'              : 'WavCaps is a dataset for testing audio captioning, where models generate textual descriptions of audio clips.',
+    'AudioCaps'            : 'AudioCaps dataset, used for generating captions from general audio events.',
+    'Clotho-AQA'           : 'Clotho dataset adapted for audio-based question answering, containing audio clips and questions.',
+    'WavCaps-QA'           : 'Question-answering test dataset derived from WavCaps, focusing on audio content.',
+    'AudioCaps-QA'         : 'AudioCaps adapted for question-answering tasks, using audio events as input for Q&A.',
+    'VoxCeleb-Accent'      : 'Test dataset for accent recognition, based on VoxCeleb, a large speaker identification dataset.',
+    'MNSC-AR-Sentence'     : 'Accent recognition based on the IMDA NSC dataset, focusing on sentence-level accents.',
+    'MNSC-AR-Dialogue'     : 'Accent recognition based on the IMDA NSC dataset, focusing on dialogue-level accents.',
+
+    'VoxCeleb-Gender': 'Test dataset for gender classification, also derived from VoxCeleb.',
+    'IEMOCAP-Gender' : 'Gender classification based on the IEMOCAP dataset.',
+    'IEMOCAP-Emotion': 'Emotion recognition test data from the IEMOCAP dataset, focusing on identifying emotions in speech.',
+    'MELD-Sentiment' : 'Sentiment recognition from speech using the MELD dataset, classifying positive, negative, or neutral sentiments.',
+    'MELD-Emotion'   : 'Emotion classification in speech using MELD, detecting specific emotions like happiness, anger, etc.',
+    'MuChoMusic'     : 'Test dataset for music understanding, from paper: MuChoMusic: Evaluating Music Understanding in Multimodal Audio-Language Models.',
+    'MNSC-PART1-ASR' : 'Speech recognition test data from the IMDA NSC project, Part 1.',
+    'MNSC-PART2-ASR' : 'Speech recognition test data from the IMDA NSC project, Part 2.',
+    'MNSC-PART3-ASR' : 'Speech recognition test data from the IMDA NSC project, Part 3.',
+    'MNSC-PART4-ASR' : 'Speech recognition test data from the IMDA NSC project, Part 4.',
+    'MNSC-PART5-ASR' : 'Speech recognition test data from the IMDA NSC project, Part 5.',
+    'MNSC-PART6-ASR' : 'Speech recognition test data from the IMDA NSC project, Part 6.',
+    'MNSC-PART3-SQA' : 'Multitak National Speech Corpus (MNSC) dataset, Question answering task, Part 3.',
+    'MNSC-PART4-SQA' : 'Multitak National Speech Corpus (MNSC) dataset, Question answering task, Part 4.',
+    'MNSC-PART5-SQA' : 'Multitak National Speech Corpus (MNSC) dataset, Question answering task, Part 5.',
+    'MNSC-PART6-SQA' : 'Multitak National Speech Corpus (MNSC) dataset, Question answering task, Part 6.',
+    'MNSC-PART3-SDS' : 'Multitak National Speech Corpus (MNSC) dataset, dialogue summarization task, Part 3.',
+    'MNSC-PART4-SDS' : 'Multitak National Speech Corpus (MNSC) dataset, dialogue summarization task, Part 4.',
+    'MNSC-PART5-SDS' : 'Multitak National Speech Corpus (MNSC) dataset, dialogue summarization task, Part 5.',
+    'MNSC-PART6-SDS' : 'Multitak National Speech Corpus (MNSC) dataset, dialogue summarization task, Part 6.',
+
+    'CNA'             : 'Under Development',
+    'IDPC'            : 'Under Development',
+    'Parliament'      : 'Under Development',
+    'UKUS-News'       : 'Under Development',
+    'Mediacorp'       : 'Under Development',
+    'IDPC-Short'      : 'Under Development',
+    'Parliament-Short': 'Under Development',
+    'UKUS-News-Short' : 'Under Development',
+    'Mediacorp-Short' : 'Under Development',
+    'YTB-ASR-Batch1'  : 'Under Development',
+    'YTB-ASR-Batch2'  : 'Under Development',
+    'SEAME-Dev-Man'   : 'Under Development',
+    'SEAME-Dev-Sge'   : 'Under Development',
+
+    'YTB-SQA-Batch1': 'Under Development',
+    'YTB-SDS-Batch1': 'Under Development',
+    'YTB-PQA-Batch1': 'Under Development',
 
-spt_datasets = {
-    'CoVoST2-EN-ID-test': 'CoVoST 2 dataset for speech translation from English to Indonesian.',
-    'CoVoST2-EN-ZH-test': 'CoVoST 2 dataset for speech translation from English to Chinese.',
-    'CoVoST2-EN-TA-test': 'CoVoST 2 dataset for speech translation from English to Tamil.',
-    'CoVoST2-ID-EN-test': 'CoVoST 2 dataset for speech translation from Indonesian to English.',
-    'CoVoST2-ZH-EN-test': 'CoVoST 2 dataset for speech translation from Chinese to English.',
-    'CoVoST2-TA-EN-test': 'CoVoST 2 dataset for speech translation from Tamil to English.'
-}
+                }
 
-cnasr_datasets = {
-    'Aishell-ASR-ZH-Test': 'ASR test dataset for Mandarin Chinese, based on the Aishell dataset.'
-}
 
-MUSIC_MCQ_DATASETS = {
-    'MuChoMusic-Test': 'Test dataset for music understanding, from paper: MuChoMusic: Evaluating Music Understanding in Multimodal Audio-Language Models.'
-}
 
-metrics = {
-    'wer'                    : 'Word Error Rate (WER), a common metric for ASR evaluation. (The lower, the better)',
-    'llama3_70b_judge_binary': 'Binary evaluation using the LLAMA3-70B model, for tasks requiring a binary outcome. (0-100 based on score 0-1)',
-    'llama3_70b_judge'       : 'General evaluation using the LLAMA3-70B model, typically scoring based on subjective judgments. (0-100 based on score 0-5)',
-    'meteor'                 : 'METEOR, a metric used for evaluating text generation, often used in translation or summarization tasks. (Sensitive to output length)',
-    'bleu'                   : 'BLEU (Bilingual Evaluation Understudy), another text generation evaluation metric commonly used in machine translation. (Sensitive to output length)',
-}
 
 metrics_info = {
     'wer'                    : 'Word Error Rate (WER) - The Lower, the better.',

app/draw_diagram.py

@@ -2,32 +2,29 @@ import streamlit as st
 import pandas as pd
 import numpy as np
 from streamlit_echarts import st_echarts
-from streamlit.components.v1 import html
-# from PIL import Image 
 from app.show_examples import *
 from app.content import *
+
 import pandas as pd
 
 from model_information import get_dataframe
-
-
-
 info_df = get_dataframe()
 
 
-def draw(folder_name, category_name, dataset_name, metrics, cus_sort=True):
+def draw(folder_name, category_name, displayname, metrics, cus_sort=True):
     
     folder = f"./results_organized/{metrics}/"
 
     # Load the results from CSV
     data_path = f'{folder}/{category_name.lower()}.csv'
     chart_data = pd.read_csv(data_path).round(3)
-    new_dataset_name = dataset_name.replace('-', '_').lower()
-    chart_data = chart_data[['Model', new_dataset_name]]
+    
+    dataset_name = displayname2datasetname[displayname]
+    chart_data = chart_data[['Model', dataset_name]]
 
     # Rename to proper display name
-    new_dataset_name = dataname_column_rename_in_table[new_dataset_name]
-    chart_data = chart_data.rename(columns=dataname_column_rename_in_table)
+    chart_data = chart_data.rename(columns=datasetname2diaplayname)
+
 
     st.markdown("""
                 <style>
@@ -52,7 +49,7 @@ def draw(folder_name, category_name, dataset_name, metrics, cus_sort=True):
                             )
     
     chart_data = chart_data[chart_data['model_show'].isin(models)]
-    chart_data = chart_data.sort_values(by=[new_dataset_name], ascending=cus_sort).dropna(axis=0)
+    chart_data = chart_data.sort_values(by=[displayname], ascending=cus_sort).dropna(axis=0)
 
     if len(chart_data) == 0: return
 
@@ -103,6 +100,19 @@ def draw(folder_name, category_name, dataset_name, metrics, cus_sort=True):
                             'IMDA-Part4-30s-ASR',
                             'IMDA-Part5-30s-ASR',
                             'IMDA-Part6-30s-ASR',
+                            'CNA',
+                            'IDPC',
+                            'Parliament',
+                            'UKUS-News',
+                            'Mediacorp',
+                            'IDPC-Short',
+                            'Parliament-Short',
+                            'UKUS-News-Short',
+                            'Mediacorp-Short',
+                            'YTB-ASR-Batch1',
+                            'YTB-ASR-Batch2',
+                            'SEAME-Dev-Man',
+                            'SEAME-Dev-Sge',
                             ]:
             
             chart_data_table = chart_data_table.sort_values(
@@ -203,7 +213,7 @@ def draw(folder_name, category_name, dataset_name, metrics, cus_sort=True):
                 "series": [{
                         "name": f"{dataset_name}",
                         "type": "bar",
-                        "data": chart_data[f'{new_dataset_name}'].tolist(),
+                        "data": chart_data[f'{displayname}'].tolist(),
                     }],
             }
             

app/pages.py

@@ -75,7 +75,7 @@ def dashboard():
     
     st.divider()
     with st.container():
-        left_co, right_co = st.columns([1, 0.7])
+        left_co, right_co = st.columns([1, 0.1])
 
         with left_co:
             st.markdown("""
@@ -88,25 +88,52 @@ def dashboard():
                             year={2024}
                             }
                         ```
+                        ```
+                        @article{wang2025advancing,
+                            title={Advancing Singlish Understanding: Bridging the Gap with Datasets and Multimodal Models},
+                            author={Wang, Bin and Zou, Xunlong and Sun, Shuo and Zhang, Wenyu and He, Yingxu and Liu, Zhuohan and Wei, Chengwei and Chen, Nancy F and Aw, AiTi},
+                            journal={arXiv preprint arXiv:2501.01034},
+                            year={2025}
+                            }
+                        ```
+                        ```
+                        @article{he2024meralion,
+                            title={MERaLiON-AudioLLM: Technical Report},
+                            author={He, Yingxu and Liu, Zhuohan and Sun, Shuo and Wang, Bin and Zhang, Wenyu and Zou, Xunlong and Chen, Nancy F and Aw, Ai Ti},
+                            journal={arXiv preprint arXiv:2412.09818},
+                            year={2024}
+                            }
+                        ```
+                        ```
+                        @article{zhang2024mowe,
+                            title={MoWE-Audio: Multitask AudioLLMs with Mixture of Weak Encoders},
+                            author={Zhang, Wenyu and Sun, Shuo and Wang, Bin and Zou, Xunlong and Liu, Zhuohan and He, Yingxu and Lin, Geyu and Chen, Nancy F and Aw, Ai Ti},
+                            journal={ICASSP},
+                            year={2025}
+                            }
+                        ```
                         """)
 
 
 
 
+
+
+
 def asr_english():
     st.title("Task: Automatic Speech Recognition - English")
     
     sum = ['Overall']
     dataset_lists = [
-                    'LibriSpeech-Test-Clean', 
-                    'LibriSpeech-Test-Other', 
-                    'Common-Voice-15-En-Test', 
-                    'Peoples-Speech-Test', 
-                    'GigaSpeech-Test', 
-                    'Earnings21-Test', 
-                    'Earnings22-Test', 
-                    'Tedlium3-Test', 
-                    'Tedlium3-Long-form-Test', 
+                    'LibriSpeech-Clean', 
+                    'LibriSpeech-Other', 
+                    'CommonVoice-15-EN', 
+                    'Peoples-Speech', 
+                    'GigaSpeech-1', 
+                    'Earnings-21', 
+                    'Earnings-22', 
+                    'TED-LIUM-3', 
+                    'TED-LIUM-3-LongForm', 
                     ]
 
     filters_levelone = sum + dataset_lists
@@ -120,7 +147,7 @@ def asr_english():
         if filter_1 in sum:
             sum_table_mulit_metrix('asr_english', ['wer'])
         else:
-            dataset_contents(asr_datsets[filter_1], metrics['wer'])
+            dataset_contents(dataset_diaplay_information[filter_1], metrics_info['wer'])
             draw('su', 'asr_english', filter_1, 'wer', cus_sort=True)
 
 
@@ -132,12 +159,12 @@ def asr_singlish():
 
     sum = ['Overall']
     dataset_lists = [
-                    'IMDA-Part1-ASR-Test', 
-                    'IMDA-Part2-ASR-Test',
-                    'IMDA-Part3-30s-ASR-Test',
-                    'IMDA-Part4-30s-ASR-Test',
-                    'IMDA-Part5-30s-ASR-Test',
-                    'IMDA-Part6-30s-ASR-Test',
+                    'MNSC-PART1-ASR', 
+                    'MNSC-PART2-ASR',
+                    'MNSC-PART3-ASR',
+                    'MNSC-PART4-ASR',
+                    'MNSC-PART5-ASR',
+                    'MNSC-PART6-ASR',
                     ]
 
     filters_levelone = sum + dataset_lists
@@ -151,7 +178,7 @@ def asr_singlish():
         if filter_1 in sum:
             sum_table_mulit_metrix('asr_singlish', ['wer'])
         else:
-            dataset_contents(singlish_asr_datasets[filter_1], metrics['wer'])
+            dataset_contents(dataset_diaplay_information[filter_1], metrics_info['wer'])
             draw('su', 'asr_singlish', filter_1, 'wer')
 
 
@@ -162,7 +189,7 @@ def asr_mandarin():
 
     sum = ['Overall']
     dataset_lists = [
-                    'Aishell-ASR-ZH-Test', 
+                    'AISHELL-ASR-ZH', 
                     ]
 
     filters_levelone = sum + dataset_lists
@@ -176,7 +203,7 @@ def asr_mandarin():
         if filter_1 in sum:
             sum_table_mulit_metrix('asr_mandarin', ['wer'])
         else:
-            dataset_contents(cnasr_datasets[filter_1], metrics['wer'])
+            dataset_contents(dataset_diaplay_information[filter_1], metrics_info['wer'])
             draw('su', 'asr_mandarin', filter_1, 'wer')
 
     
@@ -187,12 +214,12 @@ def speech_translation():
     
     sum = ['Overall']
     dataset_lists = [
-                        'CoVoST2-EN-ID-test', 
-                        'CoVoST2-EN-ZH-test',
-                        'CoVoST2-EN-TA-test', 
-                        'CoVoST2-ID-EN-test', 
-                        'CoVoST2-ZH-EN-test', 
-                        'CoVoST2-TA-EN-test']
+                        'CoVoST2-EN-ID', 
+                        'CoVoST2-EN-ZH',
+                        'CoVoST2-EN-TA', 
+                        'CoVoST2-ID-EN', 
+                        'CoVoST2-ZH-EN', 
+                        'CoVoST2-TA-EN']
 
     filters_levelone = sum + dataset_lists
     
@@ -205,7 +232,7 @@ def speech_translation():
         if filter_1 in sum:
             sum_table_mulit_metrix('st', ['bleu'])
         else:
-            dataset_contents(spt_datasets[filter_1], metrics['bleu'])
+            dataset_contents(dataset_diaplay_information[filter_1], metrics_info['bleu'])
             draw('su', 'ST', filter_1, 'bleu')
 
 
@@ -217,11 +244,11 @@ def speech_question_answering_english():
     sum = ['Overall']
 
     dataset_lists = [
-                    'CN-College-Listen-MCQ-Test',
-                    'DREAM-TTS-MCQ-Test',
-                    'SLUE-P2-SQA5-Test', 
-                    'Public-SG-Speech-QA-Test', 
-                    'Spoken-Squad-Test',
+                    'CN-College-Listen-MCQ',
+                    'DREAM-TTS-MCQ',
+                    'SLUE-P2-SQA5', 
+                    'Public-SG-Speech-QA', 
+                    'Spoken-SQuAD',
                      ]
 
     filters_levelone = sum + dataset_lists
@@ -240,7 +267,7 @@ def speech_question_answering_english():
         #    draw('su', 'SQA', filter_1, 'llama3_70b_judge')
         
         else:
-            dataset_contents(sqa_datasets[filter_1], metrics['llama3_70b_judge'])
+            dataset_contents(dataset_diaplay_information[filter_1], metrics_info['llama3_70b_judge'])
             draw('su', 'sqa_english', filter_1, 'llama3_70b_judge')
 
 
@@ -271,10 +298,39 @@ def speech_question_answering_singlish():
             sum_table_mulit_metrix('sqa_singlish', ['llama3_70b_judge'])
         
         else:
-            dataset_contents(sqa_datasets[filter_1], metrics['llama3_70b_judge'])
+            dataset_contents(dataset_diaplay_information[filter_1], metrics_info['llama3_70b_judge'])
             draw('su', 'sqa_singlish', filter_1, 'llama3_70b_judge')
 
 
+def spoken_dialogue_summarization_singlish():
+    st.title("Task: Spoken Dialogue Summarization - Singlish")
+    
+    sum = ['Overall']
+
+    dataset_lists = [
+              'MNSC-PART3-SDS', 
+              'MNSC-PART4-SDS',
+              'MNSC-PART5-SDS',
+              'MNSC-PART6-SDS',
+              ]
+
+
+    filters_levelone = sum + dataset_lists
+    
+    left, center, _, middle, right = st.columns([0.4, 0.2, 0.2, 0.2 ,0.2])
+    
+    with left: 
+        filter_1 = st.selectbox('Dataset', filters_levelone)
+
+    if filter_1:
+        if filter_1 in sum:
+            sum_table_mulit_metrix('sds_singlish', ['llama3_70b_judge'])
+        
+        else:
+            dataset_contents(dataset_diaplay_information[filter_1], metrics_info['llama3_70b_judge'])
+            draw('su', 'sds_singlish', filter_1, 'llama3_70b_judge')
+
+
 
 
 def speech_instruction():
@@ -282,8 +338,8 @@ def speech_instruction():
     
     sum = ['Overall']
 
-    dataset_lists = ['OpenHermes-Audio-Test', 
-                     'ALPACA-Audio-Test',
+    dataset_lists = ['OpenHermes-Audio', 
+                     'ALPACA-Audio',
                      ]
     
     filters_levelone = sum + dataset_lists
@@ -297,7 +353,7 @@ def speech_instruction():
         if filter_1 in sum:
             sum_table_mulit_metrix('speech_instruction', ['llama3_70b_judge'])
         else:
-            dataset_contents(si_datasets[filter_1], metrics['llama3_70b_judge'])
+            dataset_contents(dataset_diaplay_information[filter_1], metrics_info['llama3_70b_judge'])
             draw('su', 'speech_instruction', filter_1, 'llama3_70b_judge')
 
 
@@ -306,8 +362,8 @@ def speech_instruction():
 def audio_captioning():
     st.title("Task: Audio Captioning")
 
-    filters_levelone = ['WavCaps-Test', 
-                        'AudioCaps-Test',
+    filters_levelone = ['WavCaps', 
+                        'AudioCaps',
                         ]
     filters_leveltwo = ['Llama3-70b-judge', 'Meteor']
     
@@ -319,7 +375,7 @@ def audio_captioning():
         metric = st.selectbox('Metric', filters_leveltwo)
 
     if filter_1 or metric:
-        dataset_contents(ac_datasets[filter_1], metrics[metric.lower().replace('-', '_')])
+        dataset_contents(dataset_diaplay_information[filter_1], metrics_info[metric.lower().replace('-', '_')])
         draw('asu', 'audio_captioning', filter_1, metric.lower().replace('-', '_'))
 
 
@@ -330,9 +386,9 @@ def audio_scene_question_answering():
 
     sum = ['Overall']
 
-    dataset_lists = ['Clotho-AQA-Test', 
-                    'WavCaps-QA-Test', 
-                    'AudioCaps-QA-Test']
+    dataset_lists = ['Clotho-AQA', 
+                    'WavCaps-QA', 
+                    'AudioCaps-QA']
     
     filters_levelone = sum + dataset_lists
     
@@ -345,7 +401,7 @@ def audio_scene_question_answering():
         if filter_1 in sum:
             sum_table_mulit_metrix('audio_scene_question_answering', ['llama3_70b_judge'])
         else:
-            dataset_contents(asqa_datasets[filter_1], metrics['llama3_70b_judge'])
+            dataset_contents(dataset_diaplay_information[filter_1], metrics_info['llama3_70b_judge'])
             draw('asu', 'audio_scene_question_answering', filter_1, 'llama3_70b_judge')
 
 
@@ -357,9 +413,9 @@ def emotion_recognition():
     sum = ['Overall']
 
     dataset_lists = [
-                    'IEMOCAP-Emotion-Test', 
-                    'MELD-Sentiment-Test', 
-                    'MELD-Emotion-Test',
+                    'IEMOCAP-Emotion', 
+                    'MELD-Sentiment', 
+                    'MELD-Emotion',
                     ]
 
     filters_levelone = sum + dataset_lists
@@ -373,7 +429,7 @@ def emotion_recognition():
         if filter_1 in sum:
             sum_table_mulit_metrix('emotion_recognition', ['llama3_70b_judge'])
         else:
-            dataset_contents(er_datasets[filter_1], metrics['llama3_70b_judge'])
+            dataset_contents(dataset_diaplay_information[filter_1], metrics_info['llama3_70b_judge'])
             draw('vu', 'emotion_recognition', filter_1, 'llama3_70b_judge')
 
 
@@ -383,7 +439,11 @@ def accent_recognition():
     st.title("Task: Accent Recognition")
 
     sum = ['Overall']
-    dataset_lists = ['VoxCeleb-Accent-Test']
+    dataset_lists = [
+        'VoxCeleb-Accent',
+        'MNSC-AR-Sentence',
+        'MNSC-AR-Dialogue',
+        ]
 
 
     filters_levelone = sum + dataset_lists
@@ -398,7 +458,7 @@ def accent_recognition():
         if filter_1 in sum:
             sum_table_mulit_metrix('accent_recognition', ['llama3_70b_judge'])
         else:
-            dataset_contents(ar_datsets[filter_1], metrics['llama3_70b_judge'])
+            dataset_contents(dataset_diaplay_information[filter_1], metrics_info['llama3_70b_judge'])
             draw('vu', 'accent_recognition', filter_1, 'llama3_70b_judge')
 
 
@@ -409,8 +469,10 @@ def gender_recognition():
     
     sum = ['Overall']
 
-    dataset_lists =  ['VoxCeleb-Gender-Test', 
-                        'IEMOCAP-Gender-Test']
+    dataset_lists =  [
+                        'VoxCeleb-Gender', 
+                        'IEMOCAP-Gender'
+                        ]
 
     filters_levelone = sum + dataset_lists
     
@@ -423,7 +485,7 @@ def gender_recognition():
         if filter_1 in sum:
             sum_table_mulit_metrix('gender_recognition', ['llama3_70b_judge'])
         else:
-            dataset_contents(gr_datasets[filter_1], metrics['llama3_70b_judge'])
+            dataset_contents(dataset_diaplay_information[filter_1], metrics_info['llama3_70b_judge'])
             draw('vu', 'gender_recognition', filter_1, 'llama3_70b_judge')
 
 
@@ -434,7 +496,7 @@ def music_understanding():
     
     sum = ['Overall']
 
-    dataset_lists =  ['MuChoMusic-Test',
+    dataset_lists =  ['MuChoMusic',
                       ]
 
     filters_levelone = sum + dataset_lists
@@ -448,7 +510,7 @@ def music_understanding():
         if filter_1 in sum:
             sum_table_mulit_metrix('music_understanding', ['llama3_70b_judge'])
         else:
-            dataset_contents(MUSIC_MCQ_DATASETS[filter_1], metrics['llama3_70b_judge'])
+            dataset_contents(dataset_diaplay_information[filter_1], metrics_info['llama3_70b_judge'])
             draw('vu', 'music_understanding', filter_1, 'llama3_70b_judge')
 
 
@@ -457,3 +519,70 @@ def music_understanding():
 
 
 
+
+
+
+def under_development():
+    st.title("Task: Under Development")
+    
+
+    dataset_lists =  [
+                      'CNA',
+                      'IDPC',
+                      'Parliament',
+                      'UKUS-News',
+                      'Mediacorp',
+                      'IDPC-Short',
+                      'Parliament-Short',
+                      'UKUS-News-Short',
+                      'Mediacorp-Short',
+
+                      'YTB-ASR-Batch1',
+                      'YTB-ASR-Batch2',
+                      'SEAME-Dev-Man',
+                      'SEAME-Dev-Sge',
+
+                      'YTB-SQA-Batch1',
+                      'YTB-SDS-Batch1',
+                      'YTB-PQA-Batch1',
+
+                      ]
+
+    filters_levelone = dataset_lists
+    
+    left, center, _, middle, right = st.columns([0.4, 0.2, 0.2, 0.2 ,0.2])
+    
+    with left:
+        filter_1 = st.selectbox('Dataset', filters_levelone)
+    
+    dataset_contents(dataset_diaplay_information[filter_1], 'under_development')
+    
+    if filter_1 in [
+                      'CNA',
+                      'IDPC',
+                      'Parliament',
+                      'UKUS-News',
+                      'Mediacorp',
+                      'IDPC-Short',
+                      'Parliament-Short',
+                      'UKUS-News-Short',
+                      'Mediacorp-Short',
+                      'YTB-ASR-Batch1',
+                      'YTB-ASR-Batch2',
+                      'SEAME-Dev-Man',
+                      'SEAME-Dev-Sge',
+                      ]:
+        
+        draw('vu', 'under_development_wer', filter_1, 'wer')
+
+    elif filter_1 in [
+        'YTB-SQA-Batch1',
+        'YTB-SDS-Batch1',
+        'YTB-PQA-Batch1',
+        ]:
+        draw('vu', 'under_development_llama3_70b_judge', filter_1, 'llama3_70b_judge')
+
+
+
+
+

app/summarization.py

@@ -14,7 +14,6 @@ from model_information import get_dataframe
 
 info_df = get_dataframe()
 
-metrics_info = metrics_info
 
 def sum_table_mulit_metrix(task_name, metrics_lists: List[str]):
 
@@ -34,7 +33,7 @@ def sum_table_mulit_metrix(task_name, metrics_lists: List[str]):
     chart_data['Average'] = chart_data[selected_columns].mean(axis=1)
 
     # Update dataset name in table
-    chart_data = chart_data.rename(columns=dataname_column_rename_in_table)
+    chart_data = chart_data.rename(columns=datasetname2diaplayname)
     
     st.markdown("""
                 <style>
@@ -55,11 +54,9 @@ def sum_table_mulit_metrix(task_name, metrics_lists: List[str]):
     models = st.multiselect("Please choose the model", 
                             sorted(chart_data['model_show'].tolist()), 
                             default = sorted(chart_data['model_show'].tolist()),
-                            # key=f"multiselect_{task_name}_{metrics}"
                             )
     
     chart_data = chart_data[chart_data['model_show'].isin(models)].dropna(axis=0)
-    # chart_data = chart_data.sort_values(by=['Average'], ascending=True).dropna(axis=0)
 
     if len(chart_data) == 0: return