49 个 · 按相关度排序
Helps organize PyTorch model training with LightningModule and Trainer, including validation, logging, model saving, and early stopping. It also guides distributed training across multiple GPUs and settings for reduced-precision computation.
中文简介正在准备中。当前显示英语简介。
Use Hugging Face Transformers to load pretrained models, generate text, and classify text, images, or audio. Prepare model inputs, control generation settings, and fine-tune models on custom datasets with the Trainer API.
中文简介正在准备中。当前显示英语简介。
Organize PyTorch training code with LightningModule and Trainer. This skill guides training, validation, logging, model checkpoint saving, early stopping when progress stalls, and configuration for running the same model across multiple GPUs.
中文简介正在准备中。当前显示英语简介。
健身私教 — 为客户设计、指导和推进个性化训练方案的专业工艺, 有别于团课教练、物理治疗师或运动队教练: (a) 训练方案设计与周期化 (线性 / 波动 / 板块周期化 Bompa & Haff; NSCA《力量训练与体能训练精要》行业圣经; Schoenfeld 肌肥大机制; Helms/Valdez/Morgan 肌肉与力量金字塔; 共轭训练法 Simmons/Westside; Starting Strength 线性渐进 Rippetoe; Renaissance Periodization 训练量地标 Israetel); (b) 动作评估与筛查 (FMS 功能性动作筛查 Gray Cook; NASM CEx 纠正性训练连续体 抑制-拉伸-激活-整合; 关节逐级方法 Cook/Boyle; 姿势与代偿模式识别; PAR-Q+ 与 ACSM 运动前筛查指南); (c) 教学与指令 (外部 vs 内部注意焦点 Wulf; 差异化学习 Schöllhorn; 约束导向方法 Davids/Renshaw; RIR/RPE 自动调节 Zourdos/Helms; 动机访谈在健身中的应用 Rollnick/Miller); (d) 营养指导的执业边界 (宏量营养素周期化; Precision Nutrition / ISSN 立场声明的循证建议; 边界: 私教提供一般性指导, 医学营养治疗需 RD/RDN 资质; 能量可用性与 RED-S Mountjoy IOC 共识); (e) 商业与客户管理 (客户留存与方案依从; 咨询与销售伦理; 线上教练平台与规模化 Goodman PTDC; 方案交付工具 TrueCoach/Trainerize/TrainHeroic; 课时定价模型; 独立教练 vs 健身房雇员); (f) 专项方向 (运动表现 CSCS; 纠正性训练 CES; 老年健身 SFN; 产前产后; 青少年 YSCA; 适应性 / 无障碍健身; 健美 / 形体; 力量举; 耐力运动 S&C)。诚实处理: 兄弟科学 vs 循证之争, 补剂行业利益冲突, 合成代谢激素使用现状 (诚实但不推广: 匿名调查约 30% 有经验男性健身者使用过 — 业内估, Kanayama/Pope/Hudson 记录健康风险, WADA 禁用清单用于竞技语境), 体型焦虑与进食障碍风险 (男性私教肌肉变形症, 正食症倾向, NEDA 负责任语言指南), 健身房掠夺性销售 (高压推销、误导性对比照、长约锁定), 执业边界违规 (私教诊断伤病或超范围开饮食处方), 线上教练认证造假 (周末速成、未经认可的认证)。不含: 物理治疗 / 康复治疗 (持照医疗职业), 不含: 团课教练 (Les Mills/Zumba/CrossFit 分馆教练 — 相邻但不同), 不含: 精英竞技运动教练 (团队项目周期化, 不同职业路径), 不含: 营养作为主业 (临床营养学需单独资质)。 (Personal Trainer — the craft of designing, coaching, and progressing individualized exercise programs for clients, distinct from group fitness instruction, physiotherapy, or sports coaching: (a) exercise programming and periodization (linear / undulating / block periodization per Bompa & Haff; NSCA Essentials of Strength Training and Conditioning as industry bible; Schoenfeld on hypertrophy mechanisms; Helms/Valdez/Morgan Muscle & Strength Pyramids; conjugate method Simmons/Westside; Starting Strength linear progression Rippetoe; Renaissance Periodization volume landmarks Israetel); (b) movement assessment and screening (FMS Gray Cook; NASM CEx corrective exercise continuum inhibit-lengthen-activate-integrate; joint-by-joint approach Cook/Boyle; posture and compensation pattern identification; PAR-Q+ and pre-exercise screening ACSM guidelines); (c) coaching and cueing (external vs internal cueing Wulf attentional focus research; differential learning Schöllhorn; constraints-led approach Davids/Renshaw; RIR/RPE autoregulation Zourdos/Helms; motivational interviewing for adherence Rollnick/Miller adapted to fitness); (d) nutrition guidance within scope of practice (macronutrient periodization; evidence-based recommendations per Precision Nutrition / ISSN position stands; scope boundary: PT gives general guidance NOT medical nutrition therapy which requires RD/RDN credential; energy availability and RED-S awareness Mountjoy IOC consensus); (e) business and client management (client retention and programming adherence; consultation and sales ethics; online coaching platforms and scaling Goodman PTDC; programming delivery tools like TrueCoach/Trainerize/TrainHeroic; session pricing models; independent contractor vs gym employee dynamics); (f) specialization tracks (sports performance CSCS; corrective exercise CES; senior fitness SFN; pre/postnatal; youth YSCA; adaptive/inclusive fitness; bodybuilding/physique coaching; powerlifting/strength sports; endurance athlete S&C). Evidence-based practice as the core ethos: Schoenfeld/Krieger meta-analyses on volume-response, Helms systematic reviews on protein/resistance training, Brad Schoenfeld's MASS journal bridging research to practice, Menno Henselmans and Eric Helms as practitioner-researchers, Greg Nuckols Stronger By Science as translation layer. Chinese fitness industry context: rapid growth of personal training in commercial gyms (Keep/Lefit/Megafit chains), NSCA-CPT and ACE-CPT as dominant certifications alongside local CSFA (China Sport and Fitness Association) credential, '体适能' (physical fitness) as formal term, online coaching boom post-COVID, PT business models in tier 1-3 cities diverge sharply. Honest treatment of: bro science vs evidence-based d
中文简介正在准备中。当前显示原文说明。
Organize PyTorch training code with LightningModule and Trainer. Follow examples for validation, logging, model saving, and distributed training across multiple GPUs, with callbacks that add monitoring or stop training when progress stalls.
中文简介正在准备中。当前显示英语简介。
Convert existing Hugging Face Transformers Trainer or TRL SFTTrainer training code into an NVFLARE federated job using flare.patch(trainer), local validation, and job export; use when the user names Hugging Face or preliminary source inspection identifies one Hugging Face owner, and not for manual PyTorch loops, Lightning, inference-only pipelines, deployment, or experiment workflows.
中文简介正在准备中。当前显示原文说明。
K12 体育培训 (K12 / 青少年体育培训 (从业者视角) — 面向 3-18 岁青少年的校外体育培训生意，政策改革驱动的高增长赛道。覆盖: (a) 业态分层 — 少儿体适能 (3-12 岁基础运动能力/感统/协调) vs 专项培训 (篮球/游泳/跳绳/羽毛球/网球/击剑/武术等单项技能) vs 中考体育备考 (初中冲刺，按各地体育中考考纲提分); (b) 政策脊椎 — 双减 (2021 教育部「双减」: 学科类压减、非学科类规范) 把体育从「兴趣班」推成刚需，体育中考分值改革 (云南/北京/各省把分值提到 50-100 分) 制造确定性需求，《国家学生体质健康标准》+《义务教育体育与健康课程标准 (2022 版)》是教学纲领; (c) 单店模型与连锁扩张 — 坪效/师生比/续费率/转介绍率/课消，直营 vs 加盟，社区店 vs Mall 店 vs 进校 (课后服务/课后三点半); (d) 教研与师资 — 课程体系标准化 (分龄分级)、教练认证 (社会体育指导员/各单项协会教练员/中国少儿体适能等)、教练「能打 ≠ 能教」断层; (e) 运营 SOP — 招生获客 (地推/社群/试听转化)、排课与场地利用率、家长沟通与续费、安全与运动损伤防控/保险; (f) 中考体育教研 — 考纲拆解 (耐力跑/跳绳/引体/坐位体前屈/球类技能)、提分训练计划、体测数据追踪。学派分歧: 体适能通识派 vs 专项技能派、中考应试提分派 vs 长期运动素养派、直营重模型 vs 加盟轻扩张、进校 (toB/课后服务) vs 校外门店 (toC)、引进体系 (如美国 SPARK/体适能认证) vs 本土自研课程。不含: 竞技体校/专业运动员青训选材深水区、成人健身/私教 (见 personal-trainer)、学科类教培、纯体育用品零售。) Master OS — automated mastery of K12 / 青少年体育培训 (从业者视角) — 面向 3-18 岁青少年的校外体育培训生意，政策改革驱动的高增长赛道。覆盖: (a) 业态分层 — 少儿体适能 (3-12 岁基础运动能力/感统/协调) vs 专项培训 (篮球/游泳/跳绳/羽毛球/网球/击剑/武术等单项技能) vs 中考体育备考 (初中冲刺，按各地体育中考考纲提分); (b) 政策脊椎 — 双减 (2021 教育部「双减」: 学科类压减、非学科类规范) 把体育从「兴趣班」推成刚需，体育中考分值改革 (云南/北京/各省把分值提到 50-100 分) 制造确定性需求，《国家学生体质健康标准》+《义务教育体育与健康课程标准 (2022 版)》是教学纲领; (c) 单店模型与连锁扩张 — 坪效/师生比/续费率/转介绍率/课消，直营 vs 加盟，社区店 vs Mall 店 vs 进校 (课后服务/课后三点半); (d) 教研与师资 — 课程体系标准化 (分龄分级)、教练认证 (社会体育指导员/各单项协会教练员/中国少儿体适能等)、教练「能打 ≠ 能教」断层; (e) 运营 SOP — 招生获客 (地推/社群/试听转化)、排课与场地利用率、家长沟通与续费、安全与运动损伤防控/保险; (f) 中考体育教研 — 考纲拆解 (耐力跑/跳绳/引体/坐位体前屈/球类技能)、提分训练计划、体测数据追踪。学派分歧: 体适能通识派 vs 专项技能派、中考应试提分派 vs 长期运动素养派、直营重模型 vs 加盟轻扩张、进校 (toB/课后服务) vs 校外门店 (toC)、引进体系 (如美国 SPARK/体适能认证) vs 本土自研课程。不含: 竞技体校/专业运动员青训选材深水区、成人健身/私教 (见 personal-trainer)、学科类教培、纯体育用品零售。: top builders' mental models, tool stack, current workflows, jargon, and where to keep up. Trigger this skill when the user works on K12 / 青少年体育培训 (从业者视角) — 面向 3-18 岁青少年的校外体育培训生意，政策改革驱动的高增长赛道。覆盖: (a) 业态分层 — 少儿体适能 (3-12 岁基础运动能力/感统/协调) vs 专项培训 (篮球/游泳/跳绳/羽毛球/网球/击剑/武术等单项技能) vs 中考体育备考 (初中冲刺，按各地体育中考考纲提分); (b) 政策脊椎 — 双减 (2021 教育部「双减」: 学科类压减、非学科类规范) 把体育从「兴趣班」推成刚需，体育中考分值改革 (云南/北京/各省把分值提到 50-100 分) 制造确定性需求，《国家学生体质健康标准》+《义务教育体育与健康课程标准 (2022 版)》是教学纲领; (c) 单店模型与连锁扩张 — 坪效/师生比/续费率/转介绍率/课消，直营 vs 加盟，社区店 vs Mall 店 vs 进校 (课后服务/课后三点半); (d) 教研与师资 — 课程体系标准化 (分龄分级)、教练认证 (社会体育指导员/各单项协会教练员/中国少儿体适能等)、教练「能打 ≠ 能教」断层; (e) 运营 SOP — 招生获客 (地推/社群/试听转化)、排课与场地利用率、家长沟通与续费、安全与运动损伤防控/保险; (f) 中考体育教研 — 考纲拆解 (耐力跑/跳绳/引体/坐位体前屈/球类技能)、提分训练计划、体测数据追踪。学派分歧: 体适能通识派 vs 专项技能派、中考应试提分派 vs 长期运动素养派、直营重模型 vs 加盟轻扩张、进校 (toB/课后服务) vs 校外门店 (toC)、引进体系 (如美国 SPARK/体适能认证) vs 本土自研课程。不含: 竞技体校/专业运动员青训选材深水区、成人健身/私教 (见 personal-trainer)、学科类教培、纯体育用品零售。 problems and wants industry-grade thinking, tool selection, or workflow guidance. 触发词：「K12 体育培训」「青少年体育培训」「少儿体适能」「中考体育」「体育中考」
中文简介正在准备中。当前显示原文说明。
Personal trainer operations: client check-ins, workout-program delivery, progress reminders, session reminders, lead follow-up, and workout tracking. Use when an owner is a coach/personal trainer or when managing fitness clients and training routines.
中文简介正在准备中。当前显示原文说明。
Organize PyTorch models and data pipelines with PyTorch Lightning, automate training, save checkpoints, and track experiments. Includes templates and guidance for configuring training across multiple GPUs or TPUs.
中文简介正在准备中。当前显示英语简介。
Reference for the TRL (Transformer Reinforcement Learning) library codebase. Use proactively before reading or editing any file under `trl/` so you have the intended contracts and invariants in mind, not just what the current code says. Covers trainer hierarchy (SFT, DPO, GRPO, KTO), shared utility functions (selective_log_softmax, decode_and_strip_padding, padding helpers), configuration system, model wrappers, and how data flows through any TRL trainer.
中文简介正在准备中。当前显示原文说明。
Train a custom anime LoRA on the ANIMA base model with Citron's local Gradio trainer (kohya sd-scripts), <6GB VRAM, character/style LoRAs; covers setup, dataset prep, training params, and using the result in the anima-base workflow
中文简介正在准备中。当前显示原文说明。
Train object-detection, image-classification, or SAM segmentation models on Hugging Face Jobs. Use for vision fine-tuning and evaluation; use huggingface-llm-trainer for language models.
中文简介正在准备中。当前显示原文说明。
Train or fine-tune language models with TRL or Unsloth on Hugging Face Jobs, including SFT, DPO, GRPO, reward models, and GGUF conversion. Use for cloud LLM training; use huggingface-vision-trainer for vision tasks.
中文简介正在准备中。当前显示原文说明。
Build DPO training pipelines for AI alignment. TRIGGERS - Use when user needs help with ai-dpo-trainer related tasks.
中文简介正在准备中。当前显示原文说明。
Use when a task needs the judgment of a certified athletic trainer — triaging an on-field or on-court injury, deciding same-day return-to-play vs. refer-out, running a concussion sideline assessment and graduated return-to-sport progression, recognizing exertional heat stroke or another sudden-death emergency, or writing the objective findings that justify a clearance decision.
中文简介正在准备中。当前显示原文说明。
Use when a task needs the judgment of an animal trainer — designing a behavior-modification plan for a specific problem behavior, choosing a reinforcement schedule and deciding when to raise criteria, structuring session length and frequency to avoid fatigue or frustration plateaus, or setting a handler/public safety protocol around a training animal.
中文简介正在准备中。当前显示原文说明。
Prepare and launch LoRA, LoKr, or T-LoRA training through lora-scripts-next (SD-Trainer), Anima-first. Use when the user asks to train a character, style, or concept LoRA, choose parameters or repeats and epochs, start a run, or monitor training on Anima, SD1.5, SDXL, or Flux. Organize and caption the dataset, run dataset-doctor, choose a documented starting preset from image count and VRAM, show one confirmation card, then call the mikazuki API and monitor its log.
中文简介正在准备中。当前显示原文说明。
Train or fine-tune language and vision models using TRL (Transformer Reinforcement Learning) or Unsloth with Hugging Face Jobs infrastructure. Covers SFT, DPO, GRPO and reward modeling training methods, plus GGUF conversion for local deployment. Includes guidance on the TRL Jobs package, UV scripts with PEP 723 format, dataset preparation and validation, hardware selection, cost estimation, Trackio monitoring, Hub authentication, model selection/leaderboards and model persistence. Use for tasks involving cloud GPU training, GGUF conversion, or when users mention training on Hugging Face Jobs without local GPU setup.
中文简介正在准备中。当前显示原文说明。
Trains and fine-tunes vision models for object detection (D-FINE, RT-DETR v2, DETR, YOLOS), image classification (timm models — MobileNetV3, MobileViT, ResNet, ViT/DINOv3 — plus any Transformers classifier), and SAM/SAM2 segmentation using Hugging Face Transformers on Hugging Face Jobs cloud GPUs. Covers COCO-format dataset preparation, Albumentations augmentation, mAP/mAR evaluation, accuracy metrics, SAM segmentation with bbox/point prompts, DiceCE loss, hardware selection, cost estimation, Trackio monitoring, and Hub persistence. Use when users mention training object detection, image classification, SAM, SAM2, segmentation, image matting, DETR, D-FINE, RT-DETR, ViT, timm, MobileNet, ResNet, bounding box models, or fine-tuning vision models on Hugging Face Jobs.
中文简介正在准备中。当前显示原文说明。
Build DPO training pipelines for AI alignment. TRIGGERS - Use when user needs help with ai-dpo-trainer related tasks.
中文简介正在准备中。当前显示原文说明。
Build and check clinical machine learning workflows with PyHealth, from loading electronic health records, signals, or images to training and evaluation. Includes patient-level data splits, prediction task checks, model checkpoints, and medical code mapping.
中文简介正在准备中。当前显示英语简介。
Prepare GRPO reinforcement learning configurations for tasks with programmatically checkable answers, such as math, code, and structured output. Design and inspect reward functions, then select training variants when runs show instability or exploit flaws in scoring.
中文简介正在准备中。当前显示英语简介。
Guide language model training with TRL, from learning to follow instructions to matching human preferences. Covers method selection, training data, reward models that score responses, evaluation, and adjustments for memory limits or unstable training.
中文简介正在准备中。当前显示英语简介。
Help small business owners get started with Claude by guiding two tool connections and trying a task tied to their main concern. Capture business context, suggest relevant skills, and establish a weekly check-in routine.
中文简介正在准备中。当前显示英语简介。