Papers

Publications

Selected papers, benchmarks, models, and software.

2026

Report
Dynamic Semantic Tags Reduce Hallucinations in Small-LLM Post-Training
Shreyas S. Vidyarthi, Shreyas Pimpalgaonkar, Iraklis Koutrouvelis, Alexandros G. Dimakis, Maddie Daianu, Jatin Ghia, Wei Huo
Technical report · Intuit & Bespoke Labs
PDF
Benchmark
OpenThoughts-TBLite: A High-Signal Benchmark for Iterating on Terminal Agents
OpenThoughts Team
OpenThoughts blog (with Snorkel AI and Bespoke Labs)
link
arXiv
Terminal-Bench: Benchmarking Agents on Hard, Realistic Tasks in Command Line Interfaces
Mike A. Merrill, Alexander G. Shaw, Nicholas Carlini, … Shreyas Pimpalgaonkar, … Andy Konwinski, Ludwig Schmidt
arXiv preprint
link arXiv

2025

arXiv
OpenThoughts: Data Recipes for Reasoning Models
Etash Guha, Ryan Marten, … Shreyas Pimpalgaonkar, … Alexandros G. Dimakis, Ludwig Schmidt
arXiv preprint
link arXiv
ICLR
Transformers Struggle to Learn to Search
Abulhair Saparov, Srushti Pawar, Shreyas Pimpalgaonkar, Nitish Joshi, Richard Yuanzhe Pang, Vishakh Padmakumar, Seyed Mehran Kazemi, Najoung Kim, He He
ICLR 2025
link arXiv
Model
Bespoke-MiniChart-7B: Pushing the Frontiers of Open VLMs for Chart Understanding
Bespoke Labs
Bespoke Labs blog
link
Software
Curator: A Tool for Synthetic Data Creation
Ryan Marten, Trung Vu, Charlie Cheng-Jie Ji, Kartik Sharma, Shreyas Pimpalgaonkar, Alexandros G. Dimakis, Maheswaran Sathiamoorthy
Bespoke Labs, open-source software
link code
Model
Bespoke-Stratos: The Unreasonable Effectiveness of Reasoning Distillation
Bespoke Labs
Bespoke Labs blog
link

2024

Model
Triplex: A SOTA LLM for Knowledge Graph Construction
Shreyas Pimpalgaonkar, Nolan Tremelling, Owen Colegrove
Hugging Face (SciPhi.AI)
link

2021