Tag
IBM Research compares its ALTK-Evolve method to ACE, both agentic-memory systems that let LLM agents learn reusable lessons from past trajectories. ALTK-Evolve delivers individually retrievable guidelines instead of a single evolving playbook, reducing token usage while preserving non-compressed lessons with support counts.
The article analyzes the new ACE specification from the x86 Ecosystem Advisory Group, which extends Intel's AMX for AI matrix multiplication with fixed tile sizes and outer product instructions, comparing it to Arm's SME.