Karthik R. Narasimhan

Beyond Retrieval: Evidence Access and Event Aggregation in Agent Memory

Adapter Thickets: Splitting an RLVR Budget Beats Concentrating It

Majority voting over sampled completions is the workhorse of test-time scaling, and reinforcement learning with verifiable rewards (RLVR) is the workhorse for making each completion better. The standard pipeline composes the two: train one policy with RLVR, then sample it many times and vote. We show that this composit...

$\tau$-Multilingual: Benchmarking Voice Agents Across Languages

Across 4,500 full-duplex calls and five voice configurations, Spanish, Portuguese, and Hindi remain within 3.2 task-completion points of English, but Korean and Mandarin fall by 14.7 and 8.4 points, respectively.

Up and Down the Abstraction Ladder: Code-Based Skills for Language Agents

Results show that a supplied skill library can improve performance, efficiency, and learning, while retaining primitives provides flexibility when the library is insufficient, and releases CodeHack, the library of code-based skills with natural-language descriptions.

$\tau^\tau$-Bench: An Environment for End-To-End, Realistic Agent Construction

The $\tau^\tau$-bench (pronounced hyper-tau-bench), a benchmark that makes agent construction the task, is introduced to turn the work of cooperative agent building into a measurable target for coding agents.