「attention」の検索結果

Coder-agent working-file budget discipline: keep the editable working set (files you /add into writable context) under ~25k tokens, separate "read" from "edit", delegate breadth to a read-only repo-map, and drop files once edited. Use when an LLM coder-agent edits multiple files, when the working set must stay focused, or when the model starts editing the wrong file / missing targets because too much context dilutes attention. Search keywords: context window full, agent edits wrong file, too much context, /add /drop files, working file budget, context dilution, lost in the middle.

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。

Decision SOP for serving LLMs with vLLM. Covers PagedAttention mental model, quantization/parallelism/batching tradeoffs, OOM triage, and when NOT to use vLLM. Activates when a coder-agent is choosing or tuning an inference engine, debugging vLLM throughput/latency/OOM, or comparing vLLM against TGI/SGLang/TensorRT-LLM/llama.cpp.

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。

Manages attention and evidence in long agent sessions. Use for lost instructions, dropped evidence, or large multi-agent contexts.

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。

Generate scroll-stopping viral hook video prompts for Seedance 2.0 on Higgsfield. Use whenever the user wants viral content, TikTok hooks, Instagram Reels openers, YouTube Shorts, attention-grabbing video, scroll-stopper, pattern interrupt, or any short-form video designed to maximize retention and views. Triggers on any mention of viral, hook, scroll-stop, retention, views, engagement, short-form, TikTok, Reels, Shorts.

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。

Achieve aggressive v3 performance targets: 2.49x-7.47x Flash Attention speedup, 150x-12,500x search improvements, 50-75% memory reduction. Comprehensive benchmarking and optimization suite.

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。

Achieve aggressive v3 performance targets: 2.49x-7.47x Flash Attention speedup, 150x-12,500x search improvements, 50-75% memory reduction. Comprehensive benchmarking and optimization suite.

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 when the user asks to check on agents, manage their fleet, babysit panes, see what needs attention, get a status report, or when they say "check on my agents", "what needs action", "fleet status", "babysit", "brief me". Requires the control MCP server to be registered and the daemon running.

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。

Routes DINO DETR object-detection setup, COCO data validation, CUDA deformable-attention preparation, training, evaluation, inference, visualization, and bounded benchmarking tasks.

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。

Route BertViz Transformer attention visualization, neuron-view, notebook, and HTML export workflows.

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。

Kinetic typography, logo animations, and stylized motion design skill. **Triggers when:** - User wants text animations or kinetic typography - Request involves logo reveals or title sequences - Content focuses on visual impact over education - User mentions "title", "intro", "logo", "text animation", "kinetic" **Capabilities:** - Kinetic typography with dynamic text effects - Logo reveals and brand animations - Particle effects and visual flourishes - Modern motion design patterns - Attention-grabbing title sequences

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。

优化实际模型推理链路，将正确性对齐、分段 profiling、显存与数据搬运、TensorRT/ONNX/PyTorch 后端、attention/kernel、FP8/compile、缓存与少步采样、质量回归、GPU 成本和服务验收串成同一实验闭环。当用户要求推理提速、降低显存或 GPU 成本、复现模型效果、定位 GPU 利用率低、优化图像/视频/扩散模型或自托管 LLM 时使用，提供 model inference performance optimization 的脱敏案例。纯 API 调用、提示词或 token 用量优化，以及电脑卡顿，不进入本技能的模型推理实验流程。

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。

Generate attention-grabbing hooks for the first 3 seconds of videos. The hook determines if viewers stay or scroll. Creates multiple hook variations for A/B testing. Use when crafting video openings, improving retention, or creating scroll-stopping content for YouTube, TikTok, or Reels.

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。

The only memory skill that watches on its own. No database. No vectors. No manual saves. Just an LLM observer that compresses your conversations into prioritised notes, consolidates when they grow, and recovers anything missed. Five layers of redundancy, zero maintenance. ~$0.00/month (using free-tier models). While other memory skills ask you to remember to remember, this one just pays attention.

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。

Generate templated responses for common legal inquiries and identify when situations require individualized attention. Use when responding to routine legal questions — data subject requests, vendor inquiries, NDA requests, discovery holds — or when managing response templates.

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。

Think and work like an expert Digital Pathology Scientist. Use when a task calls for Digital Pathology Scientist judgment. Reasons from whole-slide pixels, pathologist-ground-truth label levels, and stain/scanner batch effects through QuPath, CLAM/TIAToolbox MIL, Macenko/Vahadane normalization, and MI-CLAIM/TRIPOD+AI standards while treating patch-level leakage, scanner-ID shortcuts, attention-on-necrosis artifacts, and inter-pathologist-kappa ceilings as first-class failure modes.

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。

Think and work like an expert Deep Learning Scientist. Use when a task calls for Deep Learning Scientist judgment. Reasons from CNN/Transformer inductive bias, Li et al. loss landscapes, grokking/mode connectivity, and Kaplan/Chinchilla scaling (~20 tokens/param); designs ResNet/ViT/DiT/MoE/FlashAttention stacks with FLOPs-matched ablations; trains AdamW+cosine/WSD via Megatron-FSDP/DeepSpeed; evaluates FID/MMLU-Pro/MMLU-CF with lm-eval decontamination and Pineau/NeurIPS reproducibility checklists.

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 this skill to plan, direct, finish, and QA provider-independent mono or stereo 180-degree, 360-degree, spherical, spatial, wide-FOV, and immersive video. It covers format choice, capture rigs, stitching, nadir and seam management, stereoscopic comfort, attention direction, stabilization, ambisonic and spatial audio, titles, captions, accessibility, edit, color, VFX, projection metadata, Vision Pro, YouTube, headset delivery, device QA, and privacy, location, and likeness rights. Do not use it for interactive XR apps, volumetric/world-model generation, game-engine experiences, or provider-specific video generation prompts.

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。

Polishes UI primitives one at a time using 2026 research-backed patterns — adds loading/icon states to buttons, replaces inline skeletons with primitives, adds secondary actions to errors, ensures 48px touch targets and visible focus indicators, applies a standard motion token ladder. Use when: 'make the UI feel more polished', 'add attention to detail', 'reduce component duplication', 'tighten the design system', 'apply craft to the frontend'.

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。

The only memory skill that watches on its own. No database. No vectors. No manual saves. Just an LLM observer that compresses your conversations into prioritised notes, consolidates when they grow, and recovers anything missed. Five layers of redundancy, zero maintenance. ~$0.00/month (using free-tier models). While other memory skills ask you to remember to remember, this one just pays attention.

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。

Audit and adapt monkey patches in vllm_metax/patch/ against a target upstream revision. Decide whether to retain, update, migrate, or remove each patch; maintain patch headers and explanations; validate changes and record evidence. Use only for compatibility work on this directory, not standalone attention backend, model, kernel, or other adaptations elsewhere in vllm_metax.

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。

Review and upgrade MetaX model support against a target vLLM revision and installed MACA components, recursively including model-dependent attention and kernels. Establish quantization and cache differences, preserve upstream structure, and distinguish shared upstream bugs from adaptation defects. Use for model support work, not standalone monkey-patch or registry audits.

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。

Review and adapt vllm_metax/registry registrations, quantization configurations, CustomOps and kernel dispatch against a target vLLM revision and installed MetaX APIs. Verify registration, inherited state, weight layouts and actual backend selection. Use for registry compatibility work, not standalone attention algorithms or monkey patches.

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。

Establish the shared environment, source/runtime correspondence, target confirmation and validation evidence for MetaX vLLM upgrades. Use as the common prerequisite of the patch, attention, registry and model upgrade skills, or for an explicitly requested MetaX environment preflight; it does not perform domain-specific adaptations.

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。

Generates MLS-ready property listing descriptions with attention-grabbing headlines, feature highlights, neighborhood context, and SEO keywords across multiple buyer-persona styles

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。