48 件 ・ 関連度順
概要と使いどころ
Train Mixture of Experts (MoE) models using DeepSpeed or HuggingFace. Use when training large-scale models with limited compute (5× cost reduction vs dense models), implementing sparse architectures like Mixtral 8x7B or DeepSeek-V3, or scaling model capacity without proportional compute increase. Covers MoE architectures, routing mechanisms, load balancing, expert parallelism, and inference optimization.
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。
Train Mixture of Experts (MoE) models using DeepSpeed or HuggingFace. Use when training large-scale models with limited compute (5× cost reduction vs dense models), implementing sparse architectures like Mixtral 8x7B or DeepSeek-V3, or scaling model capacity without proportional compute increase. Covers MoE architectures, routing mechanisms, load balancing, expert parallelism, and inference optimization.
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。
Treinar modelos de Mixture of Experts (MoE) usando DeepSpeed ou HuggingFace. Use ao treinar modelos em larga escala com computação limitada (redução de 5× em custos vs modelos densos), implementar arquiteturas esparsas como Mixtral 8x7B ou DeepSeek-V3, ou escalar capacidade de modelo sem aumento proporcional de computação. Cobre arquiteturas MoE, mecanismos de roteamento, balanceamento de carga, paralelismo de especialistas e otimização de inferência.
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。
Train Mixture of Experts (MoE) models using DeepSpeed or HuggingFace. Use when training large-scale models with limited compute (5× cost reduction vs dense models), implementing sparse architectures like Mixtral 8x7B or DeepSeek-V3, or scaling model capacity without proportional compute increase. Covers MoE architectures, routing mechanisms, load balancing, expert parallelism, and inference optimization.
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。
Provides guidance for enterprise-grade RL training using miles, a production-ready fork of slime. Use when training large MoE models with FP8/INT4, needing train-inference alignment, or requiring speculative RL for maximum throughput.
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。
Provides guidance for enterprise-grade RL training using miles, a production-ready fork of slime. Use when training large MoE models with FP8/INT4, needing train-inference alignment, or requiring speculative RL for maximum throughput.
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。
Practical guidance for training MoE VLMs in Megatron Bridge. Compares FSDP and 3D-parallel approaches, using rounded lessons from Qwen3-VL, Qwen3-Next, and other multimodal experiments.
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。
Long-context MoE training guidance for Megatron Bridge. Covers CP sizing, selective recompute, dispatcher choices, and practical patterns from DSV3, Qwen3, and Qwen3-Next long-context experiments.
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。
Evidence-gated workflow for MoE performance optimization in Megatron Bridge. Covers measurement contracts, the Three Walls framework, parallel folding, profiling, matched A/B tuning, and final validation.
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。
Choose the right MoE token dispatcher (`alltoall`, DeepEP, or HybridEP) for the hardware, EP degree, and optimization stage. Summarizes patterns from DSV3, Qwen3, Qwen3-Next, and VLM bring-up work.
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。
Validate and use MoE expert-parallel communication overlap in Megatron-Bridge, including overlap_moe_expert_parallel_comm, delay_wgrad_compute, and flex dispatcher backends such as DeepEP and HybridEP.
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。
Representative, point-in-time MoE training playbooks by hardware and model family. Use them as candidate seeds, then revalidate the exact runtime, semantics, topology, and steady-state throughput.
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。
MoE expert-parallel communication overlap in Megatron Bridge. Covers dispatch/combine overlap, flex dispatcher backends, and expert wgrad scheduling.
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。
产品营销 (PMM, Product Marketing) — 拥有「产品如何被定位 / 表达 / 发布 / 卖进市场」的从业者认知操作系统: 是产品 / 销售 / 市场 / 客户之间的连接组织, 负责把产品能力翻译成市场价值, 并把市场声音带回产品。覆盖 (a) 定位与差异化 (April Dunford 「刻意选择语境」框架 + 竞争性替代方案 + 卖价值不卖功能 + 选择「让你优势成立的市场参照系」, 对照 Ries & Trout 经典「定位是占领心智」), (b) 信息传递与叙事 (messaging 层级 / 信息屋 + 价值主张 + 收益导向 vs 战略叙事「换框架」Andy Raskin + 表达清晰度 Emma Stratton / Punchy + 初创 homepage 定位 FletchPMM), (c) 进入市场 (GTM) 策略与产品发布 (发布分级 T1/T2/T3 + 发布流程与跨职能编排 + GTM 打法 PLG vs SLG vs 混合 + 滩头细分选择), (d) 市场与竞争情报 (竞品分析 + battlecard 战卡 + 输赢分析 win/loss + 竞争赋能 Klue/Crayon/Clozd), (e) 买家与客户研究 (ICP 理想客户画像 + 买家 / 用户 persona + Jobs-to-be-Done 任务理论 Christensen/Moesta/Ulwick ODI + 客户之声 VoC + 用真实买家测信息 Wynter/Peep Laja), (f) 销售赋能 (战卡 + pitch / 销售 deck + 一页纸 + demo 叙事 + 异议处理 + 销售培训 + 内容采用率 — 你的产出只值销售实际用到的那部分), (g) 定价与打包 (价值定价 + 打包分层 + 《Monetizing Innovation》Ramanujam/Simon-Kucher + PLG 定价 — 常与产品 / 财务共担), (h) 细分与品类 (市场细分 + TAM/SAM/SOM + 技术采用生命周期与跨越鸿沟 Geoffrey Moore + 品类设计与叙事 Play Bigger / Christopher Lochhead — 稀有且昂贵, 不是默认选项), (i) 需求生成与内容协同 (campaign 信息 + 思想领导力 + 漏斗内容 TOFU/MOFU/BOFU + ABM 基于客户营销), (j) 分析师与影响者关系 (Gartner 魔力象限 + Forrester Wave + 分析师 briefing — To B 场景, 对 pay-to-play 影响诚实标注), (k) 客户营销与倡导 (参考客户 + case study + 评测站 G2/TrustRadius/Capterra 运营 + 社区), (l) PMM 运营与度量 (影响管道 + 赢率 + 发布采用率 + 信息穿透率 + 角色汇报线「向产品还是向市场」之争 + PMM 作为战略职能 vs 接单工具人陷阱); 跨 B2B SaaS (该学科重心) / B2C 消费 / 开发者 PLG / 平台产品。不含 产品管理 (PM 造产品 / 拥有 roadmap, PMM 拥有市场 — 相邻且极易混淆) / 增长 / 效果营销 / 付费投放 (平行学科) / 品牌 / 企业传播 / 公关为终点 / 需求生成为终点 (PMM 与之协同但不等同) / 平面设计 /「做 slide 和周边的团队」(PMM 抗争的接单工具人窄化) / 泛泛「市场营销」。 (Product Marketing (PMM) — the cognitive operating system of practitioners who own how a product is positioned, messaged, launched, and sold into a market: the connective tissue between product, sales, marketing, and customers, responsible for translating product capability into market value and bringing the voice of the market back into product. Covers (a) positioning & differentiation (April Dunford's deliberate-context framing, competitive alternatives, value-not-features, framing the market in which your strengths matter — vs Ries & Trout classic perception-in-the-mind positioning), (b) messaging & narrative (messaging hierarchy / house, value proposition, benefit-led vs strategic-narrative / change-the-frame messaging Andy Raskin, clarity craft Emma Stratton/Punchy, homepage & startup positioning FletchPMM), (c) go-to-market (GTM) strategy & product launches (launch tiers T1/T2/T3, launch process & cross-functional orchestration, GTM motion PLG vs SLG vs hybrid, beachhead segment selection), (d) market & competitive intelligence (competitive analysis, battlecards, win/loss analysis, competitive enablement Klue/Crayon/Clozd), (e) buyer & customer research (ICP ideal customer profile, buyer & user personas, Jobs-to-be-Done Christensen/Moesta/Ulwick ODI, voice of customer, message testing with real buyers Wynter/Peep Laja), (f) sales enablement (battlecards, pitch & sales decks, one-pagers, demo narratives, objection handling, sales training, content adoption — output is only as good as what sales actually uses), (g) pricing & packaging (value-based pricing, packaging & tiering, Monetizing Innovation Ramanujam/Simon-Kucher, PLG pricing — often shared with product/finance), (h) segmentation & category (market segmentation, TAM/SAM/SOM, technology adoption lifecycle & crossing the chasm Geoffrey Moore, category design & narrative Play Bigger / Christopher Lochhead — rare and expensive, not a default), (i) demand-gen & content partnership (campaign messaging, thought leadership, funnel content TOFU/MOFU/BOFU, ABM account-based marketing), (j) analyst & influencer relations (Gartner Magic Quadrant, Forrester Wave, analyst briefings — for B2B, with honest read on pay-to-play perception), (k) customer marketing & advocacy (references, case studies, reviews G2/TrustRadius/Capterra presence, community), (l) PMM operations & metrics (influenced pipeline, win rate, launch adoption, message pull-through, the role's reporting line product-vs-mark
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。
Design small interfering RNA and antisense oligonucleotide sequences against a transcript, and screen them for the failure modes specific to nucleic-acid drugs. Use this skill to tile a target transcript, apply positional and thermodynamic selection rules including duplex asymmetry and nearest-neighbour melting temperature, scan candidates for seed-region complementarity to off-target transcripts, and lay out a chemical modification pattern — gapmer architecture, 2'-O-methyl and 2'-MOE wings, locked nucleic acid, and phosphorothioate placement. Also trigger on siRNA, antisense oligonucleotide, ASO, gapmer, RNase H, seed region, duplex asymmetry, 2'-MOE, locked nucleic acid, phosphorothioate, or GalNAc conjugate.
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。
Fornece orientação para treinamento RL de nível empresarial usando miles, um fork pronto para produção do slime. Use ao treinar grandes modelos MoE com FP8/INT4, necessitando alinhamento treino-inferência ou exigindo RL especulativo para máxima taxa de transferência.
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。
Solves and validates single-, multi-, and many-objective optimization with pymoo, including NSGA-II, NSGA-III, MOEA/D, constraints, Pareto approximations, reference directions, and ZDT/DTLZ benchmarks for engineering and research problems.
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。
比较模型配置或架构改变对权重容量、KV 状态、前向计算和访存的影响。用于选模型、上下文或注意力与 MoE 方案。
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。
Multi-objective optimization framework. NSGA-II, NSGA-III, MOEA/D, Pareto fronts, constraint handling, benchmarks (ZDT, DTLZ), for engineering design and optimization problems.
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。
Multi-objective optimization framework. NSGA-II, NSGA-III, MOEA/D, Pareto fronts, constraint handling, benchmarks (ZDT, DTLZ), for engineering design and optimization problems.
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。
Validate and use CUDA graph capture in Megatron Bridge, including local full-iteration graphs and Transformer Engine scoped graphs for attention, MLP, and MoE modules.
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。
Convert single-node scripts to multi-node Slurm sbatch jobs and debug common multi-node failures. Covers srun-native vs uv run torch.distributed approaches, container setup, NCCL timeouts, OOM sizing for MoE models, and interactive allocation.
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。
Designs customer interviews and turns the transcripts into decisions. Produces an interview guide plus a synthesis (forces of progress, jobs, pathways, opportunities) you can paste into a doc. Use for 'customer interviews', 'discovery calls', 'JTBD', 'switch interview', 'who should I talk to', 'how many interviews', 'synthesize these transcripts', or 'why do customers churn/buy'. Draws on 140 Lenny's Podcast guests incl. Teresa Torres, Bob Moesta and Judd Antin.
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。
Incorporate moeru-ai/AUV into native desktop app control through its CLI or connected MCP for observation, targeted input, reusable operations, and verification. Use for native app automation, repeated desktop workflows, or explicit AUV requests.
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。