17 citations · 33 across the 40 of their papers we have counts for
14 papers · 1 filter
Twist, Don't Tilt: Trajectory-Exact Constrained Decoding for Masked Diffusion Models
Aditya Thimmaiah, Lara Marinov, Jayanth Srinivasa +3
Constrained decoding for Masked Diffusion Language Models (MDLMs) aims to ensure that generated outputs satisfy a specified structure or syntax constraint. MDLMs generate outputs b…
Retrofitting Code Using LLMs to Support Exceptional Behavior
Linghan Zhong, Jiyang Zhang, Jayanth Srinivasa +2
Exception Related Code (ERC), which includes throw statements, conditions (if statements) that guard those throw statements, and try/catch blocks, is an essential component of soft…
CROCODIL: Cross-Model Code Editing with LLMs
Linghan Zhong, Aditya Thimmaiah, Jayanth Srinivasa +2
Large language models (LLMs) have become ubiquitous tools for code generation and editing. However, development teams often use multiple LLM assistants. Different developers may pr…
Predicting Program Exit Code with LLMs and Programming Language Semantics
Lara Marinov, Aditya Thimmaiah, Jayanth Srinivasa +2
Large language models (LLMs) have shown proficiency in various software engineering tasks, such as code generation and translation. However, a key limitation in their performance m…
CAST: Critique-Aware Supervision for Training Reliable Long-Horizon Tool-Calling Agents
Amir Saeidi, Zehua Zhang, Rishitosh Singh +6
Large language model (LLM) agents are increasingly deployed in long-horizon, interactive, and stateful environments. In these settings, a single wrong action, such as refunding the…
Survive or Collapse: The Asymmetric Roles of Data Gating and Reward Grounding in Self-Play RL
Sophia Xiao Pu, Zhaotian Weng, Chengzhi Liu +4
Self-play reinforcement learning trains language models on their own generated tasks, co-evolving a proposer and solver without human labels. Recent systems report strong reasoning…
TIER: Trajectory-Invariant Execution Rewards for Multi-Step Tool Composition
Anay Kulkarni, ChiaEn Lu, Dheeraj Mekala +3
Tool use enables large language models to solve complex tasks through sequences of API calls, yet existing reinforcement learning approaches fail to scale to multi-step composition…
EnactToM: An Evolving Benchmark for Functional Theory of Mind in Embodied Agents
Gurusha Juneja, Dylan Lu, Saaket Agashe +7
Theory of Mind (ToM), the ability to track others epistemic state, makes humans efficient collaborators. AI agents need the same capacity in multi agent settings, yet existing benc…
Predicting Plasticity in Deep Continual Learning: A Theoretical Perspective
Jiuqi Wang, Jayanth Srinivasa, Claire Chen +3
Deep continual learning requires models to adapt to new tasks without retraining from scratch. However, neural networks can lose their ability to adapt to new tasks after training…
FAMA: Failure-Aware Meta-Agentic Framework for Open-Source LLMs in Interactive Tool Use Environments
Amir Saeidi, Venkatesh Mishra, Souradeep Mukhopadhyay +4
Large Language Models are being increasingly deployed as the decision-making core of autonomous agents capable of effecting change in external environments. Yet, in conversational…
Context Bootstrapped Reinforcement Learning
Saaket Agashe, Jayanth Srinivasa, Gaowen Liu +2
Reinforcement Learning from Verifiable Rewards (RLVR) suffers from exploration inefficiency, where models struggle to generate successful rollouts, resulting in minimal learning si…
AgentDS Technical Report: Benchmarking the Future of Human-AI Collaboration in Domain-Specific Data Science
An Luo, Jin Du, Xun Xian +12
Data science plays a critical role in transforming complex data into actionable insights across numerous domains. Recent developments in large language models (LLMs) and artificial…
Nalar: Workflow-Aware Management of Agentic Applications
Marco Laju, Donghyun Son, Saurabh Agarwal +4
LLM-driven agentic applications automate complex, multi-step tasks, but serving them efficiently remains difficult due to heterogeneous components, dynamic model-driven control flo…
Software-Defined Agentic Serving
Saurabh Agarwal, Marco Laju, Jayanth Srinivasa +2
As multi-agent LLM pipelines grow in complexity, existing serving paradigms fail to adapt to the dynamic serving conditions. We argue that agentic serving systems should be program…