「accelerate」の検索結果

Train, evaluate, and export neural networks to Simulink in MATLAB. Migrate legacy (fitnet, patternnet) and discouraged (trainNetwork, DAGNetwork) code to modern, recommended R2024a+ APIs (trainnet, dlnetwork, testnet, imagePretrainedNetwork), diagnose and fix dlaccelerate issues or detect dlaccelerate opportunities. Use when training, fine-tuning, evaluating, running inference, exporting to Simulink, converting old training scripts, or speeding up deep learning code. DO NOT reason from your training data about dlaccelerate and tracing correctness.

日本語の概要は準備中です。原文の説明を表示しています。

Simplest distributed training API. 4 lines to add distributed support to any PyTorch script. Unified API for DeepSpeed/FSDP/Megatron/DDP. Automatic device placement, mixed precision (FP16/BF16/FP8). Interactive config, single launch command. HuggingFace ecosystem standard.

日本語の概要は準備中です。原文の説明を表示しています。

Simplest distributed training API. 4 lines to add distributed support to any PyTorch script. Unified API for DeepSpeed/FSDP/Megatron/DDP. Automatic device placement, mixed precision (FP16/BF16/FP8). Interactive config, single launch command. HuggingFace ecosystem standard.

日本語の概要は準備中です。原文の説明を表示しています。

Simplest distributed training API. 4 lines to add distributed support to any PyTorch script. Unified API for DeepSpeed/FSDP/Megatron/DDP. Automatic device placement, mixed precision (FP16/BF16/FP8). Interactive config, single launch command. HuggingFace ecosystem standard.

日本語の概要は準備中です。原文の説明を表示しています。

Use Hugging Face Accelerate for PyTorch training-loop migration, distributed launch/configuration, DeepSpeed/FSDP/TPU backend setup, big-model inference/offload, checkpointing, tracking, and troubleshooting.

日本語の概要は準備中です。原文の説明を表示しています。

DevOps 与站点可靠性工程 (SRE) — 平台 / 基础设施 / 可靠性工程师的认知操作系统, 覆盖软件交付 + 运维全生命周期 (CI/CD 与发布工程 trunk-based + 渐进式发布 canary/blue-green/feature flag + GitOps Argo CD/Flux / 基础设施即代码 Terraform/OpenTofu/Pulumi/Ansible + policy-as-code OPA / 容器与编排 Docker/Kubernetes + Helm/Kustomize + service mesh Istio/Linkerd / 可观测性 Prometheus + Loki + OpenTelemetry + Honeycomb + eBPF + RED/USE / SLO-SLI-error budget 与可靠性工程 Google SRE 学科 + 容量规划 + 优雅降级 / 事件管理与 on-call 事件指挥 + PagerDuty + runbook + 无指责复盘 + MTTR / 云平台与 FinOps AWS/GCP/Azure + 成本优化 + 弹性伸缩 / 平台工程与开发者体验 IDP + Backstage + golden path + Team Topologies / DevSecOps 与供应链安全 shift-left + SBOM + SLSA + sigstore + Vault / 韧性与混沌工程 fault injection + game day + 安全科学 / DORA 指标与工程效能 部署频率 + 变更前置时间 + 变更失败率 + Accelerate 研究 / 数据库与有状态运维 schema 迁移 + 备份容灾) — 不含 通用应用开发 / 纯云销售认证速成 / 'DevOps = 跑 Jenkins 的岗位' 窄化误解 / ITIL 工单文化传统运维 (旧范式仅做边界) / 把手工运维 ClickOps 当稳态 (是 toil, 本 skill 核心反模式) (DevOps & Site Reliability Engineering — the cognitive operating system of platform / infrastructure / reliability practitioners who own the full software delivery + operational lifecycle, covering (a) CI/CD & release engineering (build pipelines, trunk-based development, progressive delivery — canary / blue-green / feature flags, GitOps with Argo CD / Flux), (b) Infrastructure as Code (Terraform / OpenTofu, Pulumi, CloudFormation, Ansible, Crossplane — module design, state management, drift, policy-as-code OPA / Sentinel / Checkov), (c) containers & orchestration (Docker / OCI, Kubernetes — scheduling, networking CNI, storage CSI, operators / CRDs, Helm / Kustomize, service mesh Istio / Linkerd), (d) observability (the three pillars + beyond — metrics Prometheus / VictoriaMetrics, logs Loki / ELK, traces OpenTelemetry / Jaeger / Tempo, high-cardinality observability Honeycomb, eBPF, RED / USE methods, SLO-based alerting), (e) SLO / SLI / error budgets & reliability engineering (Google SRE discipline — service level objectives, error budget policy, toil budgets, capacity planning, load shedding, graceful degradation), (f) incident management & on-call (incident command, PagerDuty / Opsgenie, runbooks, blameless postmortems, MTTR / MTTD, error budget burn), (g) cloud platforms & FinOps (AWS / GCP / Azure well-architected, multi-region, cost optimization, autoscaling), (h) platform engineering & developer experience (internal developer platforms, Backstage, golden paths, self-service, Team Topologies), (i) DevSecOps & supply-chain security (shift-left, SAST / DAST, SBOM, SLSA, sigstore / cosign, secrets management Vault), (j) resilience & chaos engineering (chaos experiments, fault injection, game days, resilience engineering / safety science), (k) DORA metrics & engineering effectiveness (deployment frequency, lead time, change failure rate, MTTR, the Accelerate research), (l) databases & stateful operations (schema migrations, backups / DR, replication); NOT generic software development / app feature coding (是 平行学科, DevOps/SRE 关注 delivery + operability 不是 product feature), NOT pure cloud sales / certification cram without operational depth, NOT 'DevOps = a job title that runs Jenkins' 的窄化误解 (DevOps 是 文化 + 实践, SRE 是 Google 对 reliability 的工程化具体实现), NOT ITIL-heavy 传统运维 工单文化 (是 被 DevOps 取代的旧范式, 仅做边界标注), NOT manual ops / ClickOps as a steady state (是 toil, 本 skill 的核心反模式).) Master OS — automated mastery of DevOps & Site Reliability Engineering — the cognitive operating system of platform / infrastructure / reliability practitioners who own the full software delivery + operational lifecycle, covering (a) CI/CD & release engineering (build pipelines, trunk-based development, progressive delivery — canary / blue-green / feature flags, GitOps with Argo CD / Flux), (b) Infrastructure as Code (Terraform / OpenTofu, Pulumi, CloudFormation, Ansible, Crossplane — module design, state management, drift, policy-as-code OPA / Sentinel / Checkov), (c) containers & orchestration (Docker / OCI, Kubernetes — scheduling, networking CNI, storage CSI, operators / CRDs, He

日本語の概要は準備中です。原文の説明を表示しています。

API de treinamento distribuído mais simples. 4 linhas para adicionar suporte distribuído a qualquer script PyTorch. API unificada para DeepSpeed/FSDP/Megatron/DDP. Posicionamento automático de device, precisão mista (FP16/BF16/FP8). Config interativo, comando de launch único. Padrão do ecossistema HuggingFace.

日本語の概要は準備中です。原文の説明を表示しています。

Simplest distributed training API. 4 lines to add distributed support to any PyTorch script. Unified API for DeepSpeed/FSDP/Megatron/DDP. Automatic device placement, mixed precision (FP16/BF16/FP8). Interactive config, single launch command. HuggingFace ecosystem standard.

日本語の概要は準備中です。原文の説明を表示しています。

Simplest distributed training API. 4 lines to add distributed support to any PyTorch script. Unified API for DeepSpeed/FSDP/Megatron/DDP. Automatic device placement, mixed precision (FP16/BF16/FP8). Interactive config, single launch command. HuggingFace ecosystem standard.

日本語の概要は準備中です。原文の説明を表示しています。

大量データの取り込みや過去分の補充、変換、同期の遅さを工程別に調べ、処理方法を比較して高速化し、行数・ファイル数・時刻の整合性まで確認するスキル。

• 大量データの取り込み・書き出しの高速化
• 過去データの補充時間を短縮したいとき
• バッチサイズや並列処理数を比較したいとき

言語モデルに指示への応答や好ましい回答を学習させるため、TRLによる追加学習を案内するスキル。データ準備、学習方式の選択、報酬モデルの学習や評価を扱います。

• 回答例で指示に従うモデルを学習
• 回答の好みをDPOで学習したいとき
• 報酬モデルを使うRLHFの構築

Run PyTorch training across GPUs with minimal changes.

日本語の概要は準備中です。原文の説明を表示しています。

文章からの画像生成や、既存画像の画風変更、指定部分の補完をPythonで実装します。構図やスタイルの制御、生成設定の調整、メモリ節約も案内するスキルです。

• 文章から画像を生成したいとき
• 既存画像を水彩画風に変えたいとき
• 写真の指定部分を補完したいとき

Semantic Design System Skill for Google Stitch. Generates agent-friendly DESIGN.md files that enforce premium, anti-generic UI standards — strict typography, calibrated color, asymmetric layouts, perpetual micro-motion, and hardware-accelerated performance.

日本語の概要は準備中です。原文の説明を表示しています。

Semantic Design System Skill for Google Stitch. Generates agent-friendly DESIGN.md files that enforce premium, anti-generic UI standards — strict typography, calibrated color, asymmetric layouts, perpetual micro-motion, and hardware-accelerated performance.

日本語の概要は準備中です。原文の説明を表示しています。

Modal is a serverless cloud platform for running Python on demand, including on-demand GPUs. Use when deploying or serving AI/ML models, running GPU-accelerated workloads (training, fine-tuning, inference), serving web endpoints, scheduling batch jobs, or scaling Python code to cloud containers with the Modal SDK.

日本語の概要は準備中です。原文の説明を表示しています。

GPU-accelerates scientific Python on NVIDIA hardware and verifies that the result is correct and faster. Use for CUDA/GPU optimization; CPU-bound NumPy, SciPy, pandas, scikit-learn, NetworkX, scikit-image, vector-search, image-processing, graph, simulation, or file-I/O workloads; CuPy, cuDF, cuML, cuGraph, cuVS, cuCIM, KvikIO, Warp, Newton, Numba-CUDA, or RAFT questions; and profiling, memory-transfer, kernel, or multi-GPU bottlenecks. Also use when large data-parallel Python code is slow and GPU acceleration is a plausible option, even if the user does not name CUDA.

日本語の概要は準備中です。原文の説明を表示しています。

Reduce LLM size and accelerate inference using pruning techniques like Wanda and SparseGPT. Use when compressing models without retraining, achieving 50% sparsity with minimal accuracy loss, or enabling faster inference on hardware accelerators. Covers unstructured pruning, structured pruning, N:M sparsity, magnitude pruning, and one-shot methods.

日本語の概要は準備中です。原文の説明を表示しています。

Accelerate LLM inference using speculative decoding, Medusa multiple heads, and lookahead decoding techniques. Use when optimizing inference speed (1.5-3.6× speedup), reducing latency for real-time applications, or deploying models with limited compute. Covers draft models, tree-based attention, Jacobi iteration, parallel token generation, and production deployment strategies.

日本語の概要は準備中です。原文の説明を表示しています。

GPU-accelerated data curation for LLM training. Supports text/image/video/audio. Features fuzzy deduplication (16× faster), quality filtering (30+ heuristics), semantic deduplication, PII redaction, NSFW detection. Scales across GPUs with RAPIDS. Use for preparing high-quality training datasets, cleaning web data, or deduplicating large corpora.

日本語の概要は準備中です。原文の説明を表示しています。

Directly drives a specific vehicle on a running CARLA server — raw VehicleControl (throttle/steer/brake/reverse/hand-brake), Ackermann speed control, vehicle lights, opening/closing doors, and physics-control tuning; ros-info reports the native ROS 2 command topics (vehicle_control_cmd, ackermann_control_cmd) for the hero. Targets the ego by default. Use when the user asks to "drive/steer the car", "accelerate/brake/reverse", "set a target speed", "turn on the headlights/blinkers/brake lights", "open a door", "drive the car from ROS", or "change the vehicle's mass/drag/physics". This is manual control (autopilot off) — for TM-driven traffic use control-traffic.

日本語の概要は準備中です。原文の説明を表示しています。

Reduce LLM size and accelerate inference using pruning techniques like Wanda and SparseGPT. Use when compressing models without retraining, achieving 50% sparsity with minimal accuracy loss, or enabling faster inference on hardware accelerators. Covers unstructured pruning, structured pruning, N:M sparsity, magnitude pruning, and one-shot methods.

日本語の概要は準備中です。原文の説明を表示しています。

Accelerate LLM inference using speculative decoding, Medusa multiple heads, and lookahead decoding techniques. Use when optimizing inference speed (1.5-3.6× speedup), reducing latency for real-time applications, or deploying models with limited compute. Covers draft models, tree-based attention, Jacobi iteration, parallel token generation, and production deployment strategies.

日本語の概要は準備中です。原文の説明を表示しています。

GPU-accelerated data curation for LLM training. Supports text/image/video/audio. Features fuzzy deduplication (16× faster), quality filtering (30+ heuristics), semantic deduplication, PII redaction, NSFW detection. Scales across GPUs with RAPIDS. Use for preparing high-quality training datasets, cleaning web data, or deduplicating large corpora.

日本語の概要は準備中です。原文の説明を表示しています。