27 tagged llm inference, measured the same way as everything else here.
Browse within: compilers 9gpu-computing 9inference 9llmops 9AutoGPT 8openai 8diffusion 5disaggregated-serving 5kubernetes 5omni 5routing-engine 5
Agent
Agent tracing captures request timing, token counts, worker placement, finish metadata, and replay hashes for eligible LLM requests. Requests with session identity also carry agent context, which lets analysis tools group LLM turns and tool activity into the same run.
Agent
Priority scheduling lets a client mark one request as more important than another. Dynamo exposes two related request fields.
Agent
Agent "thunderagent-program-scheduler" from ai-dynamo/dynamo, covering the problem, the scheduler, scheduler tick, tool-boundary pause/resume semantics and program lifetime.
Agent Claude Code
Agent "architect" from quantumaikr/quant.cpp, covering architect agent, 핵심 역할, 작업 원칙, 입력 and 출력.
Agent Claude Code
SIMD 최적화(ARM NEON, x86 AVX2)와 GPU 커널(CUDA, Metal) 구현. 측정 가능한 성능 개선에 집중한다.
Agent Claude Code
통합 정합성 검증 — 모듈 간 경계면에서 발생하는 불일치를 탐지한다. "존재 확인"이 아니라 "교차 비교"가 핵심이다.
Agent
How one stored agent source becomes the effective (resolved) source that is actually compiled and executed. Resolution expands FROM (inheritance) and IMPORT (textual inclusion) commitments, rewrites compact agent references into canonical URLs, and degrades gracefully when references cannot be loaded.
Agent
Compiling an agent (resolving inheritance, building model requirements, indexing KNOWLEDGE into provider-side vector stores) is expensive. The server therefore prepares agents in the background ahead of chat time and caches every derived artifact by content hash, so the first user message does not pay the indexing…
Agent
How agents move between servers and how a server's state is exported. Everything is book-centric: because a book is the single source of truth for an agent's behavior, plain .book files are the portable format.
Agent Claude Code
Use to autonomously optimize a model's AMK megakernel, runs the correctness-gated propose->eval->keep/revert loop (and the unattended autoresearch driver) and reports the measured, drift-robust speedup over AMK's own baseline.
Agent Claude Code
Senior Architect reviewer. Reviews plans and diffs for simplicity, duplication, encapsulation and abstraction. Read-only — never edits code. Use before opening a PR, or when a design decision needs a second opinion.
Agent
Refresh the Emmy model lifecycle through bounded, read-only research.
Agent
Qualify one model on the workflow-owned GPU and produce reviewed Emmy artifacts.