Datasets:

Modalities:
Audio
Text
Formats:
parquet
Size:
< 1K
Libraries:
Datasets
pandas
License:
rahular commited on
Commit
c834b3e
1 Parent(s): e520290

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +23 -22
README.md CHANGED
@@ -1,25 +1,26 @@
1
- ---
2
- dataset_info:
3
- features:
4
- - name: question
5
- dtype: string
6
- - name: lang
7
- dtype: string
8
- - name: audio
9
- dtype: audio
10
- splits:
11
- - name: train
12
- num_bytes: 218348705
13
- num_examples: 980
14
- download_size: 212931635
15
- dataset_size: 218348705
16
- configs:
17
- - config_name: default
18
- data_files:
19
- - split: train
20
- path: data/train-*
21
- pretty_name: Indic AudioLLM Evals
22
- ---
 
23
 
24
  This evaluation set contains ~100 questions in both text and audio format in Bengali, Gujarati, Hindi, Kannada, Malayalam, Marathi, Odia, Punjabi, Tamil, and Telugu. We use this dataset internally at Sarvam to evaluate the performance of our audio models. We open-source this data to enable the research community to replicate the results mentioned in our [Shuka blog](https://www.sarvam.ai/blogs/shuka-v1).
25
 
 
1
+ ---
2
+ dataset_info:
3
+ features:
4
+ - name: question
5
+ dtype: string
6
+ - name: lang
7
+ dtype: string
8
+ - name: audio
9
+ dtype: audio
10
+ splits:
11
+ - name: train
12
+ num_bytes: 218348705
13
+ num_examples: 980
14
+ download_size: 212931635
15
+ dataset_size: 218348705
16
+ configs:
17
+ - config_name: default
18
+ data_files:
19
+ - split: train
20
+ path: data/train-*
21
+ pretty_name: Indic AudioLLM Evals
22
+ license: apache-2.0
23
+ ---
24
 
25
  This evaluation set contains ~100 questions in both text and audio format in Bengali, Gujarati, Hindi, Kannada, Malayalam, Marathi, Odia, Punjabi, Tamil, and Telugu. We use this dataset internally at Sarvam to evaluate the performance of our audio models. We open-source this data to enable the research community to replicate the results mentioned in our [Shuka blog](https://www.sarvam.ai/blogs/shuka-v1).
26