Scholar

Sachin Gururangan

Google Scholar ID: CJIKhNIAAAAJ

Anthropic

Natural Language ProcessingMachine Learning

Citations & Impact

All-time

Citations

19,789

H-index

i10-index

Publications

Co-authors

list available

Contact

Publications

20 items

Browse publications on Google Scholar (top-right) ↗

Resume (English only)

Academic Achievements

Published multiple influential papers in 2025 on LLM reasoning and training, including:
- "Rethinking Thinking Tokens: LLMs as Improvement Operators": Introduced the PDR framework, achieving +11% on AIME 2024 and +9% on AIME 2025
- "Compute as Teacher: Turning Inference Compute Into Reference-Free Supervision": Proposed CaT method, yielding ~+27% on MATH-500 and +12% on HealthBench, with RL enhancements reaching +33% and +30%
- "Diversity-driven Data Selection for Language Model Tuning through Sparse Autoencoder": Leveraged sparse autoencoders for diversity-aware data selection, improving model performance while reducing compute costs
- "BTS: Harmonizing Specialized Experts into a Generalist LLM": Introduced the BTS training paradigm for merging domain experts into a generalist LLM
Recipient of the 2022 Bloomberg PhD Fellowship

Co-authors

8 total