Gaurav Srivastava (@ctrlgaurav) profile picture

Gaurav Srivastava

@ctrlgaurav

AI Research Intern @Dell Office of the CTO. Graduate Student @VT_CS

 Austin, TX, USA
linkedin.com/in/gaurav-srivastava-gk/
Joined March 2025
  • Tweets8
  • Following71
  • Followers8
LLMs can do complex math—but struggle with sorting numbers? We announce LLMThinkBench to evaluate how LLMs reason (or overthink) with basic math reasoning. Huge thanks to Dr. Xuan Wang (@xwang174), Dr. Tu Vu (@tuvllms), Dr. Christine Julien (@drcjulien), Aafiya, and Sriram!
1
1
1
4
531
Hindsight is a collaboration between @Vectorizeio @washingtonpost and researchers @SanghaniCtrVT (Gaurav Srivastava @xwang174 @profnaren). It is based on this research: arxiv.org/abs/2512.12818
Hindsight is 20/20: Building Agent Memory that Retains, Recalls,...

Hindsight is 20/20: Building Agent Memory that Retains, Recalls,...

Agent memory has been touted as a dimension of growth for LLM-based applications, enabling agents that can accumulate experience, adapt across sessions, and move beyond single-shot question...

arxiv.org
VentureBeat (@VentureBeat) profile pictureVentureBeat
@VentureBeat16 Dec 2025
With 91% accuracy, open source Hindsight agentic memory provides 20/20 vision for AI agents stuck on failing RAG venturebeat.com/ai/with-91-a…
With 91% accuracy, open source Hindsight agentic memory provides 20/20 vision for AI agents stuck...

With 91% accuracy, open source Hindsight agentic memory provides 20/20 vision for AI agents stuck...

Vectorize.io's Hindsight system achieves 91.4% accuracy on LongMemEval, offering enterprises a robust alternative to RAG for consistent AI agent performance.

venturebeat.com
3
2
1,382
BeyondBench: Contamination-Resistant Evaluation of Reasoning in Language Models Main proceedings @ctrlgaurav @aafiya_hussain @zhenyubvt @pitre_priya Meng Lu @xwang174 (all VT) w/Swastik Roy, Morteza Ziyadi @Amazon AGI arxiv.org/abs/2509.24210
BeyondBench: Contamination-Resistant Evaluation of Reasoning in...

BeyondBench: Contamination-Resistant Evaluation of Reasoning in...

Evaluating language models fairly is increasingly difficult as static benchmarks risk contamination by training data, obscuring whether models truly reason or recall. We introduce BeyondBench, an...

arxiv.org
1
1
2
80
The International Conference on Learning Representations (@iclr_co) has convened in Rio de Janeiro, Brazil, and runs through April 27. A number of @SanghaniCtrVT faculty and students are presenting their research. See 🧵in no particular order.
1
1
1
2
519
Gaurav Srivastava retweeted
Pau Labarta Bajo (@paulabartabajo_) profile picture
Pau Labarta Bajo
@paulabartabajo_Feb 8
Building agentic systems using Small Language Models (as opposed to large cloud models) is faster and cheaper. Here's a library a very promising library that can help you github.com/ctrl-gaurav/effGe…
GitHub - ctrl-gaurav/effGen: [ICML 2026] effGen: Enabling Small Language Models as Capable Autono...

GitHub - ctrl-gaurav/effGen: [ICML 2026] effGen: Enabling Small Language Models as Capable Autono...

[ICML 2026] effGen: Enabling Small Language Models as Capable Autonomous Agents - ctrl-gaurav/effGen

github.com
7
14
92
5,286
Gaurav Srivastava retweeted
Xuan Wang (@xwang174) profile picture
Xuan Wang@xwang17428 Jul 2025
Welcome to join our 𝗔𝗖𝗟 𝟮𝟬𝟮𝟱 𝗕𝗶𝗿𝗱𝘀-𝗼𝗳-𝗮-𝗙𝗲𝗮𝘁𝗵𝗲𝗿 (𝗕𝗼𝗙) 𝘀𝗲𝘀𝘀𝗶𝗼𝗻: 𝗕𝗿𝗶𝗱𝗴𝗶𝗻𝗴 𝗛𝘂𝗺𝗮𝗻 𝗦𝘁𝘂𝗱𝘆 𝗮𝗻𝗱 𝗟𝗟𝗠 𝗔𝗴𝗲𝗻𝘁𝘀 𝗳𝗼𝗿 𝗦𝗼𝗰𝗶𝗮𝗹 𝗦𝗶𝗺𝘂𝗹𝗮𝘁𝗶𝗼𝗻! 🌐 Website: lnkd.in/eq5n3vTz @aclmeeting @priyapitre @ctrlgaurav
Image from Xuan Wang (@xwang174)
1
7
16
1,578
Gaurav Srivastava retweeted
Xuan Wang (@xwang174) profile picture
Xuan Wang@xwang17411 Jul 2025
@priyapitre @ctrlgaurav
1
218
🧠 Fully open-source and customizable 👇 Try it: 📦pip install llmthinkbench (pypi.org/project/llmthinkben…) 🔗 github.com/ctrl-gaurav/LLMTh… 🆕 v0.1.5 is live with better error handling, fallback support, and multi-GPU parallelization! Stay tuned—benchmark results on 50+ LLMs coming soon!
llmthinkbench

llmthinkbench

A framework for evaluating overthinking and basic reasoning capabilities of Large Language Models

pypi.org
2
72
Load more