Vishruth Veerendranath

AI @ Snowflake | ML/NLP Master’s @LTIatCMU

[Vishruth Veerendranath @viishruth]

@nealchopra Congrats!!

[Kshitish Ghate @GhateKshitish]

🚨New paper: Reward Models (RMs) are used to align LLMs, but can they be steered toward user-specific value/style preferences?
With EVALUESTEER, we find even the best RMs we tested exhibit their own value/style biases, and are unable to align with a user >25% of the time. 🧵

[Devansh Jain @devanshrjain]

Thrilled to launch Prompt Adaptation, a state-of-the-art agentic system to automate prompt engineering 🚀

[Zora Wang @ZhiruoW]

Excited to share that AWM has been accepted at #ICML2025 🥳
Check out our online memory-adaptive agent if you haven't! 🔗https://t.co/Ktrs2odilY

[Zora Wang @ZhiruoW]

Cannot attend #ICLR2025 in person (will be NAACL and Stanford soon!), but do check out 👇
▪️Apr 27: "Exploring the Pre-conditions for Memory-Learning Agents" led by @viishruth and Vishwa Shah, at SSI-FM workshop
▪️Apr 28: our @DL4Code workshop with a fantastic line of works & speakers!
▪️Apr 28: @dan_fried's talk at the DL4C about "Inducing Functions to Improve LLM Agents"

[Zora Wang @ZhiruoW]

Meet ASI: Agent Skill Induction
A framework for online programmatic skill learning — no offline data, no training.
🧠 Build reusable skills during test
📈 +23.5% success, +15.3% efficiency
🌐 Scales to long-horizon tasks, transfers across websites

Let's dive in! 🧵

[Rajhans Samdani @rajhans_samdani]

Another banger from my group that tbh raises more questions about creating data agents then answers.

Here's the core issue:

When creating an agent to query structured & unstructured data for business insights, how do you describe these data tools?

Let me elaborate 🧵👇

[Rajhans Samdani @rajhans_samdani]

🧵 1/ Recent hype suggests long-context LLMs remove the need for retrieval in RAG pipelines—"just put all your docs in context." We tested this theory rigorously in finance, focusing on SEC filings.

Spoiler: Retrieval & chunking strategies still dominate.

[Snowflake @Snowflake]

Snowflake Cortex Agents, now in public preview!

Cortex Agents orchestrates across structured and unstructured data for accurate AI-driven decisions from within the secure Snowflake perimeter; Cortex Agents use Cortex Analyst (now in GA) and Cortex Search as tools.

@AnthropicAI's Claude 3.5 Sonnet is used by Cortex Agents to deliver accurate, efficient and governed data insights at scale. All the details: https://t.co/jSGMFmmD3N

[Graham Neubig @gneubig]

How far are we from having competent AI co-workers that can perform tasks as varied as software development, project management, administration, and data science?

In our new paper, we introduce TheAgentCompany, a benchmark for AI agents on consequential real-world tasks.

[Zora Wang @ZhiruoW]

Excited to co-organized the DL4C workshop at @iclr_conf'25. Check out our call for papers and submit your interesting codegen paper! 😉

[Arjun Choudhry @Arjun_7m]

Excited to share TimeSeriesExam for systematic evaluation of time series reasoning capabilities of LLMs. Think your LLM can reason on time series concepts? Take it for a spin on the TimeSeriesExam! Now publicly available on HuggingFace :)

[Daniel Campos @spacemanidol]

🚀 I am thrilled to introduce @SnowflakeDB 's Arctic Embed 2.0 embedding models! 2.0 offers high-quality multilingual performance with all the greatness of our prior embedding models (MRL, Apache-2 license, great English retrieval, inference efficiency) https://t.co/hEcd0niVyr🌍

[Aurick Qiao @aurickq]

We are excited to share SwiftKV, our recent work at @SnowflakeDB AI Research! SwiftKV reduces the pre-fill compute for enterprise LLM inference by up to 2x, resulting in higher serving throughput for input-heavy workloads. 🧵

[Matthew Leavitt @leavittron]

Tired: Bringing up politics at Thanksgiving

Wired: Bringing up @datologyai’s new text curation results at Thanksgiving

That’s right, we applied our data curation pipeline to text pretraining data and the results are hot enough to roast a 🦃

🧵

[Matthew Leavitt @leavittron]

🧵We’ve spent the last few months at @datologyai building a state-of-the-art data curation pipeline and I’m SO excited to share our first results: we curated image-text pretraining data and massively improved CLIP model quality, training speed, and inference efficiency 🔥🔥🔥

[Language Technologies Institute | @CarnegieMellon @LTIatCMU]

ECCO: Can We Improve Model-Generated Code Efficiency Without Sacrificing Functional Correctness?

by @s_waghjale, @viishruth, @ZhiruoW, and @dan_fried

Session: NLP Applications 1, Session 02, 11:00-12:30

https://t.co/YpAEBhpUNf

[Language Technologies Institute | @CarnegieMellon @LTIatCMU]

As we prepare for EMNLP 2024 (@emnlpmeeting), we're thrilled to congratulate all of the LTI researchers with accepted papers at this year's conference. In all, 32 papers with LTI authors were accepted! Read about them all here:
https://t.co/o77FizuTE4

[Zora Wang @ZhiruoW]

Sad to miss #EMNLP2024 but do check out our paper "ECCO: Can We Improve Model-Generated Code Efficiency Without Sacrificing Functional Correctness?" presented by @viishruth and Siddhant
Tuesday 11-12:30 at Poster Session 02‼️

[Vishruth Veerendranath @viishruth]

I’m attending #EMNLP2024 in Miami from 11-16th Nov to present ECCO on Tuesday 🏖️

Looking forward to meeting folks and chatting more about code generation and LLM agents!

[Vishruth Veerendranath @viishruth]

Can current code LMs generate sufficiently efficient programs? 🤔
More importantly, Can these LMs improve code efficiency without sacrificing correctness?

Check out ECCO, our code-gen benchmark for correctness-preserving program optimizations!
🧵 1/n