'acl' 검색 결과

host / 辅助段优化模块（L3）：把**非 DiT、非解码段**的固定开销落成可复验的优化—— ① 交付与搬运（mp4/图片编码、worker→API 通路、落盘、异步化）； ② 装载与预热（权重加载、编译与图下发预热、镜像/容器预热）。 当用户说“mp4 编码/传输/落盘慢”“交付这段占了不少时间”“模型加载慢”“首启编译太久” “固定开销占比高”“每个请求都要重新编码/重新写盘”时使用。 **入口条件**：占比已达门限（瓶颈标签 `非DiT-host段`；标签枚举、门限与测点口径的单一真源在 model-auto-optimization/references/bottleneck-labels.md，本技能只消费不定义）。 near-miss（看似相关但不属本技能）： - **并发 / 吞吐 / 批处理调度 / 队列深度与限流** → **不属本技能**（会改变所有性能口径与 SLO 语义， 未来独立立项；本技能只在**不改变请求并发模型**的前提下压固定开销）； - VAE/TAE 解码段（计算或通信）→ `vae-opt`；DiT 计算 → `dit-perf-opt`； 多卡并行 / 通信掩盖 → `dit-parallel-opt`； - **装库 / 下载权重 / 构建镜像 / 创建容器** → `env-install`（本技能只优化装载与预热**过程本身**）； - 编译与图下发的**机制本体**（pattern / Inductor / NPUGraph / 图 pool / 缓存条目）→ `pattern-dev` / `aclgraph-dev`（本技能不另做一套缓存）； - 产物一致性与精度判据 → `accuracy-gate`；收益口径 → `perf-gate`； - 瓶颈未定位 / 没有占比锚点 → `model-auto-optimization`。

한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.

PyTorch Inductor pattern matcher 机制开发：扩展本仓 compile 能力（default/Inductor 后端）， 或扩展三方框架自带的类似能力——Pattern 创建/注册/调试、GraphPatternEntry 手动改图、 Copy 算子（InplaceCopy/ViewCopy）消减的机制与全生命周期，含 Phase 1–7 出口证据纪律。 当用户需要写/改/调试 pattern、处理融合不命中、做 Copy/InplaceCopy 消减或注册 replacement 时 使用；即使用户只说"pattern 不生效""copy 算子变多了""compile 没融合"而未说全称，也应触发 （触发词：pattern / pattern matcher / register_replacement / GraphPatternEntry / compile 融合 / Copy 消减）。 边界：批量下发（图 capture/replay）见 aclgraph-dev；算子本体开发见 operator-dev； 框架侧特性使能/补齐见 framework-integration。 由 dev-workflow 的 pattern/compile 开发场景路由触发，亦由 model-auto-optimization 的 S1 kernel 融合（接入/compile）替换场景指向（框架侧使能不支持、需新增/适配 pattern 时）。

한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.

DiT 计算模块（L3）：把**已定位的 DiT 计算瓶颈**落成特性级选档与实施—— 量化档（W8A16 / W4A16 / W8A8 系列 / W4A4 / MXFP8 / FA 量化）、稀疏（rf_v2 / ada_bsa）、 缓存（DiTCache / AttentionCache / 时间步优化）、编译启用（MindieSDBackend / Pattern 融合 / ACLGraph） 的**开不开、开哪一档、怎么开、怎么复验**；依据是 `docs/zh/features/*`（特性真源）+ framework-integration/references/framework-support-matrix.md（支持状态）。 即使用户只说"这个模型怎么加速""量化/稀疏/Cache 怎么选怎么开""要不要开量化、开哪一档""这个档位开了有没有效果" 而未提 profiling，也应触发。 **入口条件**：瓶颈点已明确（用户带一句实测锚点，或编排层交付标签）时由域入口 `performance-optimization` 按标签分发到本技能；**瓶颈未明（"怎么加速 / 跑通 / 采 profile"）先走 `model-auto-optimization` 定位**，不在本技能内做占比分析。 near-miss：多卡并行形态 / 通信掩盖 / TP·offload 选型 → `dit-parallel-opt`；VAE 解码段与 host 固定开销 → 各自模块（VAE / host）；单算子实现级实测选型（mindie_bench）→ `benchmark-dev`； 需要新增 pattern / 算子才能落地本档 → `pattern-dev` / `operator-dev`；框架侧开关与使能验证 → `framework-integration`；量化器位级契约与精度对齐（编码公式 / 舍入 / scale 粒度）→ `quantization-dev`；精度验收判据 → `accuracy-gate`；数字入库口径 → `perf-gate`。

한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.

Analyze bonded agent systems, escrow workflows, slashing rules, oracle trust, Sybil cost, and front-running risk in protocols where agents post bonds or stake reputation. Use for collateral sizing, griefing analysis, dispute design, and bonded-agent security reviews. NOT for smart contract auditing, DeFi tokenomics, blockchain consensus design, or trading strategy.

한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.

Conversation patterns and interaction protocols for multi-agent systems. Covers request/response, pub/sub, blackboard, delegation chains, debate, critique, consensus, fan-out/fan-in, supervisor-worker, and peer negotiation. Deep analysis of AutoGen conversation patterns, CrewAI delegation, LangGraph state passing, and FIPA-ACL performatives. Teaches how to design what agents say to each other and in what order. Activate on: "agent conversation", "agent protocol", "multi-agent debate", "agent delegation", "supervisor worker pattern", "agent voting", "consensus protocol", "fan-out fan-in", "agent negotiation", "blackboard pattern", "agent dialogue", "conversation topology", "agent handoff". NOT for: wire format or serialization (use agent-interchange-formats), orchestration infrastructure (use agentic-infrastructure-2026), single agent behavior (use agentic-patterns).

한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.

Auto-activate for pytest_databases, Docker DB fixtures, PostgreSQL/pgvector/ParadeDB, MySQL/MariaDB, Oracle/SQL Server, CockroachDB, MongoDB, Redis/Valkey, BigQuery/Spanner, MinIO, or RustFS. Not for mocked DBs.

한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.

Auto-activate for polyfactory, ModelFactory, DataclassFactory, MsgspecFactory, AttrsFactory, Use, register_fixture, pytest plugin, __random_seed__, or coverage(). Not for production seeding.

한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.

Apply GEODE's measured architecture, module and symbol naming, Python type/class, persisted schema, test, site, and versioning conventions. Use when adding or moving modules, choosing dataclass/Pydantic/Protocol/ABC forms, naming classes or files, designing stored data, reviewing convention compliance, or deciding compatibility and release version treatment.

한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.

Simulate transcription factor perturbation effects on cell state in silico with CellOracle and Dynamo, and predict transcriptional responses to genetic perturbations with GEARS, scGen, and CPA. Covers the direction-not-magnitude principle, local-linear validity, the GRN/velocity error it inherits, baseline discipline (mean and additive baselines), and the validation gap. Use when predicting TF knockout or overexpression effects, ranking driver TFs for fate transitions, or planning perturbation experiments. For GRN construction see multiomics-grn; for experimental Perturb-seq see single-cell/perturb-seq.

한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.

Build enhancer-driven gene regulatory networks (eGRNs) by integrating single-cell RNA-seq and ATAC-seq using SCENIC+, CellOracle base GRNs, Pando, FigR, DIRECT-NET, TRIPOD, and scMEGA. Covers the accessibility-defines-enhancers principle, peak-to-gene linking and its cell-composition confound, the paired-vs-unpaired decision, and TF-region-gene eRegulon triplets. Use when analyzing 10x multiome or paired/unpaired scRNA+scATAC to infer cis-regulatory GRNs. For RNA-only regulons see scenic-regulons; for in silico TF perturbation see perturbation-simulation.

한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.

Maps between gene identifier systems (Ensembl, Entrez, HGNC symbol, UniProt, RefSeq, MANE) using AnnotationDbi, biomaRt, mygene, pyensembl, and Ensembl REST. Encodes Ensembl version stripping with GENCODE _PAR_Y preservation, the Ziemann 2016 Excel autocorrect debacle and Bruford 2020 HGNC renames (SEPT*->SEPTIN*, MARCH*->MARCHF*, MARC*->MTARC*, DEC1->DELEC1), OCT4/POU5F1 alias resolution, biomaRt archive endpoints for release pinning, the `filters` (plural) gotcha, MANE Select for clinical reporting, cross-species orthology via Ensembl Compara / OMA / OrthoDB, and tx2gene construction for tximport. Use when converting gene IDs across systems, handling renamed symbols, building tx2gene, pinning to a specific Ensembl release for reproducibility, or mapping cross-species orthologs.

한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.

Detect introgression and admixture between species or populations using Dsuite (Malinsky 2021 fast D-statistics), Patterson's D / ABBA-BABA test (Green 2010; Durand 2011), f4-ratio and f-branch statistic (Malinsky 2018), TreeMix (Pickrell & Pritchard 2012), HyDe (Blischak 2018), QuIBL (Edelman 2019), sprime (Browning 2018), Twisst (Martin 2017), PhyloNet (Than 2008) for explicit phylogenetic networks, and qpAdm / qpGraph (Patterson 2012). Distinguish introgression from incomplete lineage sorting (ILS), ancestral structure, ghost-lineage admixture, and rate variation. Use when testing inter-species gene flow, dating admixture events, identifying introgressed segments, building phylogenetic networks for reticulate evolution, or applying the ABBAclustering (Koppetsch-Malinsky-Matschiner 2024) framework for divergent-species gene flow.

한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.

Call protein-RNA binding sites from CLIP-seq BAM with CLIPper, PureCLIP, Skipper, Piranha, omniCLIP, CTK, CLAM, or Paraclu. Use when choosing between coverage-based, HMM-based, beta-binomial window-based, and crosslink-site-based peak callers; applying ENCODE eCLIP thresholds (log2 IP/SMInput >= 3, -log10 p >= 3); deciding when SMInput is mandatory; or reconciling peak-set discordance between callers for the same RBP.

한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.

Reads, validates, and prepares CDISC SDTM and ADaM clinical trial data for analysis. Covers SDTM domain joins (DM, AE, EX, VS, LB, DS), ADaM architecture (ADSL, BDS, OCCDS, ADTTE) with traceability, treatment-emergent AE conventions, baseline derivation, SUPPQUAL/NSV handling, Define-XML 2.1, and Pinnacle 21 / CORE validation. Use when working with clinical trial datasets in CDISC SDTM/ADaM format, preparing analysis-ready data, or validating for regulatory submission.

한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.

Evaluates patient-specific factors such as lifestyle, motivation, and barriers that may affect successful CGM use. Triggered when clinicians discuss patient’s readiness for CGM or identify obstacles to consistent sensor wear.

한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.

Recommends fludrocortisone 100 µg daily for children with primary adrenal insufficiency and confirmed aldosterone deficiency; for infants under 12 months, adds sodium chloride supplementation. Triggered by PAI with aldosterone deficiency, evidenced by salt craving, hyponatremia, hyperkalemia, or elevated renin with low aldosterone.

한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.

Decompose an MVP vision into a sprint manifest — a prioritized feature list where every feature carries explicit done-criteria (a done-record reference, named edge tests, and a wiring test). Produces the sprint manifest that `sprint-oracle` checks for completeness. Use when planning a sprint of parallel work (Slice A of the sprint-mode epic). Not a workflow phase; the manifest it writes is the completeness contract a sprint is held to.

한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.

Oracle-bound spec-review check that every structured Rollout prerequisite binds to an enforcement-type acceptance criterion. A prerequisite whose `enforced-by` is missing, dangling, or points at a non-enforcement AC is a BLOCKER (hard-blocks implementation entry via the checker fan-out verdict); a prerequisite left in free prose is ADVISORY. Read-only. Runs in the spec-review fan-out before implementation.

한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.

Use when building creative browser demos with @chenglou/pretext — DOM-free text layout for ASCII art, typographic flow around obstacles, text-as-geometry games, kinetic typography, and text-powered generative art. Produces single-file HTML demos by default.

한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.

Use when an idea becomes a Pixar-style animation prompt — character sheet, scene, image, video; emotional beat, want, obstacle. Triggers 'Pixar prompt', 'animated scene'. Live-action → video-director.

한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.

Connect to any database — Cloud SQL, PostgreSQL, Snowflake, Databricks, Athena, Presto, or Oracle.

한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.

Universal SQL performance optimization assistant for comprehensive query tuning, indexing strategies, and database performance analysis across all SQL databases (MySQL, PostgreSQL, SQL Server, Oracle). Provides execution plan analysis, pagination optimization, batch operations, and performance monitoring guidance. Use when a query is slow, an index strategy is unclear, or database performance needs analysis.

한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.

Reviews SQL for security (injection, privileges), performance, maintainability, and quality across MySQL, PostgreSQL, SQL Server, Oracle, and other dialects. Use when reviewing a query, migration, stored procedure, or the SQL in a pull request.

한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.

Use Exasol JDBC-based virtual schemas for federated read-only access to external databases such as PostgreSQL, Oracle, MySQL, SQL Server, and DB2. Covers database-source adapter selection, `CREATE VIRTUAL SCHEMA`, `ALTER VIRTUAL SCHEMA`, `EXPLAIN VIRTUAL`, metadata refresh, and query-side troubleshooting.

한국어 요약은 준비 중입니다. 원문 설명을 표시하고 있습니다.