AI Paper Atlas

Research map / Large language models / Language, safety and interpretability

BhashaKritika: Building Synthetic Pretraining Data at Scale for Indic Languages

Guduru Manoj, Neel Prabhanjan Rachamalla, Ashish Kulkarni, Gautam Rajeev, Jay Piplodiya, Arul Menezes, Shaharukh Khan, Souvik Rana, Manya Sah, Chandra Khatri, Shubham Agarwal

AAAI 2026 · Large language models › Language, safety and interpretability · approach: LLM pruning and scaling data · 2 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
use synthetic BhashaKritika Indic data instead of Web data in annealingLLaMA-3.2 1B · Indic Sentimentaccuracy: 0.617 -> 0.631Improved
use synthetic data instead of Web in annealingLLaMA-3.2 1B · GSM8Kaccuracy: 0.111 -> 0.120Improved
use synthetic data instead of Web in annealingLLaMA-3.2 1B · Indic XParaphraseaccuracy: 0.782 -> 0.736Hurt
use synthetic data instead of Web in annealingLLaMA-3.2 1B · Indic XNLIaccuracy: 0.433 -> 0.403Hurt
use synthetic data instead of Web in annealingLLaMA-3.2 1B · Hellaswag, MMLU, OpenbookQAaccuracy: 0.483 -> 0.482 (Hellaswag); 0.412 -> 0.408 (MMLU)No clear effect
continue pretraining on synthetic data instead of repeated Web epochsLLaMA-3.2 1B from scratch · Indic XNLIaccuracy: 0.330 -> 0.339Improved
continue pretraining on synthetic data instead of repeated Web epochsLLaMA-3.2 1B from scratch · ARC Challenge Indicaccuracy: 0.213 -> 0.210No clear effect
add English prompt to Indic source document (vs Indic prompt) (ablation)document-grounded generation · Gujarati discard ratediscard rate (%): 24.15 -> 15.85Improved
use Indic document instead of English document with English prompt (ablation)document-grounded generation · Gujarati discard ratediscard rate (%): 17.45 -> 15.85Improved
use Indic prompt instead of English prompt for Indic documents (ablation)document-grounded generation · Bengali discard ratediscard rate (%): 0.47 -> 0.33Improved
add matching document grounding to English persona (ablation)persona-based generation · discard ratediscard rate (%): 1.93% -> 3.50%Hurt
add random document grounding to English persona (ablation)persona-based generation · discard ratediscard rate (%): 1.93% -> 14.43%Hurt

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy