IndicIFBench

IndicIFBench is a translation of IFBench, a benchmark for verifiable instruction following, into the 22 scheduled languages of India. Each prompt combines a task with one or more instructions whose compliance can be checked automatically (for example "include the keyword river twice" or "answer in exactly 3 paragraphs"). The instructions, and the parameters the checkers use, are translated together so that every prompt stays verifiable in its language.

6,336 prompts: 288 per language.

Code	Language	Script	Prompts
as	Assamese	Bengali-Assamese	288
bn	Bengali	Bengali	288
brx	Bodo	Devanagari	288
doi	Dogri	Devanagari	288
gu	Gujarati	Gujarati	288
hi	Hindi	Devanagari	288
kn	Kannada	Kannada	288
kok	Konkani	Devanagari	288
ks	Kashmiri	Perso-Arabic	288
mai	Maithili	Devanagari	288
ml	Malayalam	Malayalam	288
mni	Manipuri	Meetei Mayek	288
mr	Marathi	Devanagari	288
ne	Nepali	Devanagari	288
or	Odia	Odia	288
pa	Punjabi	Gurmukhi	288
sa	Sanskrit	Devanagari	288
sat	Santali	Ol Chiki	288
sd	Sindhi	Devanagari	288
ta	Tamil	Tamil	288
te	Telugu	Telugu	288
ur	Urdu	Perso-Arabic	288

Loading

from datasets import load_dataset

ds = load_dataset("ai4bharat/IndicIFBench", "hi", split="train")

There is one configuration per language code.

Data fields

The schema follows IFBench:

Field	Description
key	Row identifier, the same as in English IFBench, so the same prompt can be compared across languages.
prompt	The translated prompt.
instruction_id_list	The IFBench instruction IDs that apply to the prompt.
kwargs	One dictionary of checker parameters per instruction, in the same order. Keywords, answer options and texts to repeat or overlap with are translated and match the prompt exactly.

Two additions to the IFBench format:

• kwargs[i]._lang: translated material the checker needs for this language. fixed holds translated fixed strings (template labels such as "My Answer:", CSV column names, the sentence to repeat, the answer options' "or", the European capitals) and names holds the transliterated person names of count:person_names.
• response_language:xx: a few prompts ask for the answer in another language (they translate a passage into Russian, or rewrite it in modern English). These rows carry an extra scored instruction, response_language:ru (keys 52, 181) or response_language:en (keys 117, 206, 208, 240, 248), with empty parameters, which checks the answer's language.

How it was built

1. Selection. 288 of the 300 English IFBench prompts. 12 were left out because their instructions cannot carry over to Indic languages: title case and sentences starting with A to Z (Indic scripts have no letter case or A to Z), starting with a verb, and an English alphabetical-order task.
2. Translation with Gemini. Keywords, answer options, person names, texts to repeat or overlap with, and fixed strings were translated once and synchronised between the prompt and kwargs. Numbers stay in ASCII digits. Code, formulas, URLs and established acronyms stay as in the source.
3. Adaptation. Instructions defined on English letters use the target script: the alphabet runs from the script's first to its last letter, and instructions on letters, vowels, consonant clusters and syllables state the unit they count in that script. Prompts that ask to translate, rewrite or check a given text keep that text in its original language (keys 52, 115, 117, 181, 206, 208, 240, 243, 248).
4. Quality control. Every translation was reviewed by an LLM judge, flagged rows were repaired by an LLM corrector, and automatic consistency checks were run on all rows (every parameter appears in its prompt, keywords and names as whole words, keyword counts match the English, no letters from another script, no non-ASCII digits). A few rows were repaired by hand.

Status: machine-translated with automatic quality control. Human verification by native speakers is in progress; this card will be updated when it is complete.

Evaluation

The data uses IFBench's instruction IDs and parameter format. Many original IFBench checkers assume English text (word lists, sentence splitting, letters, keyword boundaries), so scoring Indic responses requires language-aware checkers that read _lang and apply each script's rules.

License

ODC-BY-1.0, the license of the IFBench data, which this dataset translates. It is intended for research and educational use in accordance with Ai2's Responsible Use Guidelines.

Citation

Please cite IFBench, which this dataset is based on:

@misc{pyatkin2025generalizing,
   title={Generalizing Verifiable Instruction Following},
   author={Valentina Pyatkin and Saumya Malik and Victoria Graf and Hamish Ivison and Shengyi Huang and Pradeep Dasigi and Nathan Lambert and Hannaneh Hajishirzi},
   journal={Advances in Neural Information Processing Systems},
   volume={38},
   year={2025}
}