evaluation benchmarks 1local language models 1multi-agent systems 1prompt engineering 1self-refinement 1
From the 1 of 2 linked papers with an AI index.
2 papers
cs.LG2026
Two Calls Beat Five Agents: Evaluating Multi-Agent Pipelines Against Self-Refinement for Local Language Models
Ashish Prajapati, Om Mohite
The paper compares a five‑role multi‑agent LLM pipeline with a simpler two‑call self‑refinement approach on a local 7B model, finding that communication format and implementation d…
cs.AR2026
FusionML: Prefill, Not Decode - Mechanism and Boundaries of CPU+GPU Co-Execution on Unified-Memory Apple Silicon
Om Mohite
Apple-Silicon SoCs share CPU, GPU, and Neural Engine over one unified memory system, raising the question of whether transformer inference can be accelerated by splitting single op…