l3cube-pune/IndicGuard
IndicGuard: Multilingual Safety Guard Model and Dataset for Indic Languages Dataset Summary IndicGuard is a large-scale, culturally nuanced multilingual safety dataset designed for training and evaluating LLM guardrail models across ten major Indic languages. The dataset comprises approximately 334,160 samples (33,416 per language), spanning training, validation, test, and refusal splits. The dataset is constructed to address a critical gap in safety alignment… See the full description on the dataset page: https://huggingface.co/datasets/l3cube-pune/IndicGuard.
| Author / organization | l3cube-pune |
|---|---|
| License (as declared) | cc-by-4.0 |
| Downloads | 198 (as reported by Hugging Face) |
| Likes | 3 |
| Created | 21 June 2026 |
| Last updated at source | 19 July 2026 |
Recent changes
- 16 Oct 2026 License: apache-2.0 → cc-by-4.0
- 16 Oct 2026 Description: none → IndicGuard: Multilingual Safety Guard Model and Dataset for Indic Languages Dataset Summary IndicGuard is a large-scale, culturally nuanced multilingual safety dataset designed for training and evaluating LLM guardrail models across ten maj
- 16 Oct 2026 Topics added task_categories:text-classification, guardrail, jailbreak-detection, llm-alignment removed peft, safetensors, lora, sft, transformers, trl, unsloth, text-generation, conversational, hi, mr, bn, ta, te, kn, ml, gu, pa, or, dataset:l3cube-pune/IndicGuard
- 16 Oct 2026 Task: text-generation → none
Changes detected by Silytics between refreshes of the source.
More from l3cube-pune
Silytics shows public metadata from Hugging Face and links to the original. Information may be incomplete or out of date. Last refreshed 16 October 2026. Licenses and terms of the original apply to the work itself. Not financial, legal or professional advice.