update tamil page

app/content.py

@@ -212,6 +212,7 @@ wer_displayname2datasetname = {
     'YouTube ASR: Malay with English Prompt': 'ytb_asr_batch3_malay',
     'YouTube ASR: Chinese with English Prompt': 'ytb_asr_batch3_chinese',
     'YouTube ASR: Tamil with English Prompt': 'ytb_asr_batch3_tamil',
+    'YouTube ASR: Tamil with English Prompt V2': 'ytb_asr_batch3_tamil_v2',
 
     'YouTube ASR: Malay with Malay Prompt': 'ytb_asr_batch3_ms_ms_prompt',
     'YouTube ASR: Chinese with Chinese Prompt': 'ytb_asr_batch3_zh_zh_prompt',
@@ -384,6 +385,8 @@ dataset_diaplay_information = {
     
     'YouTube ASR: Tamil with English Prompt': 'YouTube Evaluation Dataset for ASR Task: <br> This dataset contains Tamil and some Tamil-English codeswitch audio clips, featuring with English prompts. <br> It includes approximately 2.44 hours of audio, with individual clips ranging from 30 seconds to 324 seconds in length.',
     
+    'YouTube ASR: Tamil with English Prompt V2': 'YouTube Evaluation Dataset for ASR Task: <br> This dataset contains Tamil and some Tamil-English codeswitch audio clips, featuring with English prompts. <br> It includes approximately 2.44 hours of audio, with individual clips ranging from 30 seconds to 324 seconds in length.',
+
     'YouTube ASR Translation: Malay2English': 'YouTube Evaluation Dataset for ASR Task: <br> The audio of dataset is same as <i>YouTube ASR: Malay<i>',
     
     # 'YouTube ASR Translation: Chinese2English': 'YouTube Evaluation Dataset for ASR Task: <br> The audio of dataset is same as <i>YouTube ASR: Chinese<i>',

app/summarization.py

@@ -29,6 +29,9 @@ def sum_table_mulit_metrix(task_name, metrics_lists: List[str]):
             chart_data = pd.merge(chart_data, one_chart_data, on='Model', how='outer')
 
     selected_columns = [i for i in chart_data.columns if i != 'Model']
+    # TODO: temp code. delete this after ytb tamil vs is fully tested. 
+    _columns_to_exclude_from_average = ["ytb_asr_batch3_tamil_v2"]
+    selected_columns = [col for col in selected_columns if col not in _columns_to_exclude_from_average]
     chart_data['Average'] = chart_data[selected_columns].mean(axis=1)
 
     # Update dataset name in table
@@ -54,8 +57,11 @@ def sum_table_mulit_metrix(task_name, metrics_lists: List[str]):
                             sorted(chart_data['model_show'].tolist()), 
                             default = sorted(chart_data['model_show'].tolist()),
                             )
-    
-    chart_data = chart_data[chart_data['model_show'].isin(models)].dropna(axis=0)
+    # TODO: delete this after ytb tamil v2 is fully utilized. 
+    if task_name == 'asr_tamil':
+        chart_data = chart_data[chart_data['model_show'].isin(models)]
+    else:
+        chart_data = chart_data[chart_data['model_show'].isin(models)].dropna(axis=0)
 
     if len(chart_data) == 0: return
 

model_information.py

@@ -14,7 +14,7 @@ data['Link'].append('https://arxiv.org/abs/2511.09690')
 
 data['Original Name'].append('MERaLiON-ASR-dev-1215')
 data['Proper Display Name'].append('🌟 API: MERaLiON-ASR-dev-1215')
-data['Link'].append("")
+data['Link'].append(None)
 
 data['Original Name'].append('SALMONN_7B')
 data['Proper Display Name'].append('Fusion: SALMONN-7B')
@@ -95,6 +95,6 @@ def get_dataframe():
     Returns a DataFrame with the data and drops rows with missing values.
     """
     df = pd.DataFrame(data)
-    return df.dropna(axis=0)
+    return df
 
 

results_organized/wer/asr_tamil.csv

@@ -1,18 +1,18 @@
-Model,commonvoice_17_ta_asr,fleurs_tamil_ta_30_asr,ytb_asr_batch3_tamil
-MERaLiON-AudioLLM-Whisper-SEA-LION,0.5284951114826634,0.4624736472241743,0.6929759165018962
-MERaLiON-AudioLLM-v2-2b,0.1385300804387941,0.1432185523541813,0.7504943113675407
-MERaLiON-AudioLLM-v2-9b,0.1559177057102368,0.1608573436401967,0.6644679264853651
-MERaLiON-AudioLLM-v2-9b-asr,0.1287122656417262,0.1383345045678145,0.5467894071504975
-Qwen2.5-Omni-3B,0.8307319012713203,1.653935347856641,1.460763022268322
-Qwen2.5-Omni-7B,0.8465494917777076,0.8666549543218552,1.3615441962983372
-SALMONN_7B,1.4272941368377052,1.507519325368939,0.985267900554277
-SeaLLMs-Audio-7B,1.2968793010286783,2.061876317638791,3.617451622313701
-cascade_whisper_large_v2_gemma2_9b_cpt_sea_lionv3_instruct,0.2380539724938065,0.2724525650035137,0.9665002755178114
-cascade_whisper_large_v3_llama_3_8b_instruct,0.2440435531721838,0.283977512297962,0.8976532365239376
-hy_whisper_local_cs,0.3179371374392121,0.3311314125087842,0.8339924151567211
-phi_4_multimodal_instruct,1.1784589191228196,1.7016514406184118,2.750056724255292
-whisper_large_v3,0.2448438631011245,0.2314476458186929,0.8481572720495284
-MERaLiON-SpeechEncoder2-ASR-CTC,0.1442,0.1632,0.6578
-Omnilingual-ASR-7B,0.3144055763521363,0.1062680115273775,0.867817443980474
-Fusion: Omnilingual-LLM-ASR-7B[with language code],0.3144055763521363,0.1062680115273775,0.8675527848026818
-MERaLiON-ASR-dev-1215,0.12422135451181095,0.13475052705551652,0.49745551197692134
+Model,commonvoice_17_ta_asr,fleurs_tamil_ta_30_asr,ytb_asr_batch3_tamil,ytb_asr_batch3_tamil_v2
+MERaLiON-AudioLLM-Whisper-SEA-LION,0.5284951114826634,0.4624736472241743,0.6929759165018962,
+MERaLiON-AudioLLM-v2-2b,0.1385300804387941,0.1432185523541813,0.7504943113675407,
+MERaLiON-AudioLLM-v2-9b,0.1559177057102368,0.1608573436401967,0.6644679264853651,0.5082255864938114
+MERaLiON-AudioLLM-v2-9b-asr,0.1287122656417262,0.1383345045678145,0.5467894071504975,0.3790156916225034
+Qwen2.5-Omni-3B,0.8307319012713203,1.653935347856641,1.460763022268322,
+Qwen2.5-Omni-7B,0.8465494917777076,0.8666549543218552,1.3615441962983372,
+SALMONN_7B,1.4272941368377052,1.507519325368939,0.985267900554277,
+SeaLLMs-Audio-7B,1.2968793010286783,2.061876317638791,3.617451622313701,
+cascade_whisper_large_v2_gemma2_9b_cpt_sea_lionv3_instruct,0.2380539724938065,0.2724525650035137,0.9665002755178114,
+cascade_whisper_large_v3_llama_3_8b_instruct,0.2440435531721838,0.283977512297962,0.8976532365239376,
+hy_whisper_local_cs,0.3179371374392121,0.3311314125087842,0.8339924151567211,
+phi_4_multimodal_instruct,1.1784589191228196,1.7016514406184118,2.750056724255292,
+whisper_large_v3,0.2448438631011245,0.2314476458186929,0.8481572720495284,
+MERaLiON-SpeechEncoder2-ASR-CTC,0.1442,0.1632,0.6578,
+Omnilingual-ASR-7B,0.3144055763521363,0.1062680115273775,0.867817443980474,
+Fusion: Omnilingual-LLM-ASR-7B[with language code],0.3144055763521363,0.1062680115273775,0.8675527848026818,
+MERaLiON-ASR-dev-1215,0.12422135451181095,0.13475052705551652,0.49745551197692134,0.3112948609504739