Tag: IBM Research

Showing 1-2 of 2

Apr 18, 20268 min readAgent PlaybookIntermediate180 min build

VAKRA benchmark: reproducible execution traces for diagnosing multi-step agent tool use

Guides running VAKRA's runnable benchmark—8,000+ local APIs across 62 domains—to record full execution traces, reproduce common multi‑step agent failures, and guide focused fixes.

VAKRA agents benchmarking tool-use failure-modes

+3 more

TutorialsFrance

Open

Apr 08, 20268 min readAgent PlaybookIntermediate120 min build

ALTK‑Evolve: Distilling Agent Transcripts into Reusable Guidelines for Long‑Term Memory

How ALTK‑Evolve converts agent interaction traces into short, human‑reviewed guidelines and injects only relevant rules at decision time to improve reliability on multi‑step tasks.

ALTK-Evolve agents long-term memory on-the-job learning Hugging Face

+3 more

IBM Research ReAct CUGA