Karthik Narasimhan
YOU?
Author Swipe
View article: Persona-Driven Benchmarking for Generalizable and Human-Aware Artificial General Intelligence
Persona-Driven Benchmarking for Generalizable and Human-Aware Artificial General Intelligence Open
This research paper, "Persona-Driven Benchmarking for Generalizable and Human-Aware Artificial General Intelligence," proposes a novel architectural solution to transition current Large Language Models (LLMs) from sophisticated pattern-mat…
View article: Probing AI Safety with Source Code
Probing AI Safety with Source Code Open
Large language models (LLMs) have become ubiquitous, interfacing with humans in numerous safety-critical applications. This necessitates improving capabilities, but importantly coupled with greater safety measures to align these models wit…
View article: RLHF Deciphered: A Critical Analysis of Reinforcement Learning from Human Feedback for LLMs
RLHF Deciphered: A Critical Analysis of Reinforcement Learning from Human Feedback for LLMs Open
A significant challenge in training large language models (LLMs) as effective assistants is aligning them with human preferences. Reinforcement learning from human feedback (RLHF) has emerged as a promising solution. However, our understan…
View article: Agent Context Protocols Enhance Collective Inference
Agent Context Protocols Enhance Collective Inference Open
AI agents have become increasingly adept at complex tasks such as coding, reasoning, and multimodal understanding. However, building generalist systems requires moving beyond individual agents to collective inference -- a paradigm where mu…
View article: Contextual Experience Replay for Self-Improvement of Language Agents
Contextual Experience Replay for Self-Improvement of Language Agents Open
View article: An Annotated Dataset of Errors in Premodern Greek and Baselines for Detecting Them
An Annotated Dataset of Errors in Premodern Greek and Baselines for Detecting Them Open
View article: PersonaGym: Evaluating Persona Agents and LLMs
PersonaGym: Evaluating Persona Agents and LLMs Open
View article: LoRA Soups: Merging LoRAs for Practical Skill Composition Tasks
LoRA Soups: Merging LoRAs for Practical Skill Composition Tasks Open
Low-Rank Adaptation (LoRA) is a popular technique for parameter-efficient fine-tuning of Large Language Models (LLMs). We study how different LoRA modules can be merged to achieve skill composition -- testing the performance of the merged …
View article: SWE-bench Multimodal: Do AI Systems Generalize to Visual Software Domains?
SWE-bench Multimodal: Do AI Systems Generalize to Visual Software Domains? Open
Autonomous systems for software engineering are now capable of fixing bugs and developing features. These systems are commonly evaluated on SWE-bench (Jimenez et al., 2024a), which assesses their ability to solve software issues from GitHu…
View article: EnIGMA: Interactive Tools Substantially Assist LM Agents in Finding Security Vulnerabilities
EnIGMA: Interactive Tools Substantially Assist LM Agents in Finding Security Vulnerabilities Open
Although language model (LM) agents have demonstrated increased performance in multiple domains, including coding and web-browsing, their success in cybersecurity has been limited. We present EnIGMA, an LM agent for autonomously solving Ca…
View article: LLMs are Superior Feedback Providers: Bootstrapping Reasoning for Lie Detection with Self-Generated Feedback
LLMs are Superior Feedback Providers: Bootstrapping Reasoning for Lie Detection with Self-Generated Feedback Open
Large Language Models (LLMs) excel at generating human-like dialogues and comprehending text. However, understanding the subtleties of complex exchanges in language remains a challenge. We propose a bootstrapping framework that leverages s…
View article: ShieldGemma: Generative AI Content Moderation Based on Gemma
ShieldGemma: Generative AI Content Moderation Based on Gemma Open
We present ShieldGemma, a comprehensive suite of LLM-based safety content moderation models built upon Gemma2. These models provide robust, state-of-the-art predictions of safety risks across key harm types (sexually explicit, dangerous co…
View article: PersonaGym: Evaluating Persona Agents and LLMs
PersonaGym: Evaluating Persona Agents and LLMs Open
Persona agents, which are LLM agents conditioned to act according to an assigned persona, enable contextually rich and user aligned interactions across domains like education and healthcare. However, evaluating how faithfully these agents …
View article: $τ$-bench: A Benchmark for Tool-Agent-User Interaction in Real-World Domains
$τ$-bench: A Benchmark for Tool-Agent-User Interaction in Real-World Domains Open
Existing benchmarks do not test language agents on their interaction with human users or ability to follow domain-specific rules, both of which are vital for deploying them in real world applications. We propose $τ$-bench, a benchmark emul…
View article: SWE-agent: Agent-Computer Interfaces Enable Automated Software Engineering
SWE-agent: Agent-Computer Interfaces Enable Automated Software Engineering Open
Language model (LM) agents are increasingly being used to automate complicated tasks in digital environments. Just as humans benefit from powerful software applications, such as integrated development environments, for complex tasks like s…
View article: Can Language Models Solve Olympiad Programming?
Can Language Models Solve Olympiad Programming? Open
Computing olympiads contain some of the most challenging problems for humans, requiring complex algorithmic reasoning, puzzle solving, in addition to generating efficient code. However, it has been understudied as a domain to evaluate lang…
View article: RLHF Deciphered: A Critical Analysis of Reinforcement Learning from Human Feedback for LLMs
RLHF Deciphered: A Critical Analysis of Reinforcement Learning from Human Feedback for LLMs Open
State-of-the-art large language models (LLMs) have become indispensable tools for various tasks. However, training LLMs to serve as effective assistants for humans requires careful consideration. A promising approach is reinforcement learn…
View article: Language-Guided World Models: A Model-Based Approach to AI Control
Language-Guided World Models: A Model-Based Approach to AI Control Open
This paper introduces the concept of Language-Guided World Models (LWMs) -- probabilistic models that can simulate environments by reading texts. Agents equipped with these models provide humans with more extensive and efficient control, a…
View article: GEO: Generative Engine Optimization
GEO: Generative Engine Optimization Open
The advent of large language models (LLMs) has ushered in a new paradigm of search engines that use generative models to gather and summarize information to answer user queries. This emerging technology, which we formalize under the unifie…
View article: QualEval: Qualitative Evaluation for Model Improvement
QualEval: Qualitative Evaluation for Model Improvement Open
Quantitative evaluation metrics have traditionally been pivotal in gauging the advancements of artificial intelligence systems, including large language models (LLMs). However, these metrics have inherent limitations. Given the intricate n…
View article: Progressively Efficient Learning
Progressively Efficient Learning Open
Assistant AI agents should be capable of rapidly acquiring novel skills and adapting to new user preferences. Traditional frameworks like imitation learning and reinforcement learning do not facilitate this capability because they support …
View article: SWE-bench: Can Language Models Resolve Real-World GitHub Issues?
SWE-bench: Can Language Models Resolve Real-World GitHub Issues? Open
Language models have outpaced our ability to evaluate them effectively, but for their future development it is essential to study the frontier of their capabilities. We find real-world software engineering to be a rich, sustainable, and ch…
View article: FireAct: Toward Language Agent Fine-tuning
FireAct: Toward Language Agent Fine-tuning Open
Recent efforts have augmented language models (LMs) with external tools or environments, leading to the development of language agents that can reason and act. However, most of these agents rely on few-shot prompting techniques with off-th…
View article: Cognitive Architectures for Language Agents
Cognitive Architectures for Language Agents Open
Recent efforts have augmented large language models (LLMs) with external resources (e.g., the Internet) or internal control flows (e.g., prompt chaining) for tasks requiring grounding or reasoning, leading to a new class of language agents…
View article: Scaling Laws for Imitation Learning in Single-Agent Games
Scaling Laws for Imitation Learning in Single-Agent Games Open
Imitation Learning (IL) is one of the most widely used methods in machine learning. Yet, many works find it is often unable to fully recover the underlying expert behavior, even in constrained environments like single-agent games. However,…
View article: COLLIE: Systematic Construction of Constrained Text Generation Tasks
COLLIE: Systematic Construction of Constrained Text Generation Tasks Open
Text generation under constraints have seen increasing interests in natural language processing, especially with the rapidly improving capabilities of large language models. However, existing benchmarks for constrained generation usually f…
View article: InstructEval: Systematic Evaluation of Instruction Selection Methods
InstructEval: Systematic Evaluation of Instruction Selection Methods Open
In-context learning (ICL) performs tasks by prompting a large language model (LLM) using an instruction and a small set of annotated examples called demonstrations. Recent work has shown that precise details of the inputs used in the ICL p…
View article: InterCode: Standardizing and Benchmarking Interactive Coding with Execution Feedback
InterCode: Standardizing and Benchmarking Interactive Coding with Execution Feedback Open
Humans write code in a fundamentally interactive manner and rely on constant execution feedback to correct errors, resolve ambiguities, and decompose tasks. While LLMs have recently exhibited promising coding capabilities, current coding b…
View article: PruMUX: Augmenting Data Multiplexing with Model Compression
PruMUX: Augmenting Data Multiplexing with Model Compression Open
As language models increase in size by the day, methods for efficient inference are critical to leveraging their capabilities for various applications. Prior work has investigated techniques like model pruning, knowledge distillation, and …
View article: Referral Augmentation for Zero-Shot Information Retrieval
Referral Augmentation for Zero-Shot Information Retrieval Open
We propose Referral-Augmented Retrieval (RAR), a simple technique that concatenates document indices with referrals, i.e. text from other documents that cite or link to the given document, to provide significant performance gains for zero-…