“deepseek”的搜索结果

Use when writing or reviewing tests for DeepSeek Harness plugins, external DSH plugin packages, or package changes in the deepseek-harness repository. Select the minimum sufficient levels from unit tests, coverage, real-API end-to-end tests, snapshots, Web tests, real compositions, and built-artifact smoke tests. For Harness version migrations, use this Skill's built-in seven-touchpoint test workflow and validate the exact target-version runtime.

中文简介正在准备中。当前显示原文说明。

Use when creating a DeepSeek Harness plugin, choosing public names for a new external DSH plugin, validating a dsh-plugin.naming.json manifest, checking reviewed central registrations for known conflicts, or creating a workspace package inside the deepseek-harness repository. Covers the full workflow from repository-mode and plugin-form selection through separate offline naming validation, optional online registry lookup, and package validation. Routes tool, LLM adapter, hook, service, and configuration forms to their corresponding references while separating upstream-monorepo rules from external-package rules. For an existing plugin crossing Harness versions, use this Skill's built-in version-adaptation workflow before implementing against the exact target contract.

中文简介正在准备中。当前显示原文说明。

DeepSeek Harness 开发文档离线 Wiki，检索导航涵盖的插件开发、工具、配置、生命周期、服务、事件、LLM 适配器和 Cordis 教程。用户询问这些开发主题时使用；不用于日常产品使用、部署运维或完整 API 参考。

中文简介正在准备中。当前显示原文说明。

公众号写稿｜长文写作｜文章润色｜改写续写 — 公众号长文 AI 写作，从话题或提纲生成完整初稿，支持改写、续写、润色、开头结尾优化，可调 DeepSeek / GPT / Claude 或由 Agent 代写。面向自媒体作者、公众号运营、品牌文案。触发词（**单独触发仅限对已有稿子的修改**）：「改写」「润色」「续写」「续一段」「往下写」「接着这段写」「重写开头」「改结尾」「调整语气」「这段润色下」「把这段改活泼点」「优化用词」「用 GPT 重写」「用 DeepSeek 重写」。新写一篇请走 aws-wechat-article-main（main 内部会调用本 skill 生成初稿）；需要多环节串联（写+审+排+配图+发）也走 main。

中文简介正在准备中。当前显示原文说明。

Build with DeepSeek's models — efficient open-weight reasoning models and low-cost API inference.

中文简介正在准备中。当前显示原文说明。

Build with DeepSeek Coder — open code models with strong reasoning and long-context code understanding.

中文简介正在准备中。当前显示原文说明。

Gate resumption of any multi-day LLM batch-scoring campaign that calls an unpinned model alias (deepseek-chat, gpt-*-latest, gemini-*-preview, any provider alias without a pinned version). Use when: (1) resuming a paused or credit-exhausted scoring run days after its last chunk, (2) topping up credits to finish a campaign, (3) extending a cached scoring pipeline with new items. Prevents silently splicing two model versions or serving revisions into one measure. Verified 2026-07-16: for $0.30 caught a serving-revision drift WITHIN DeepSeek v4-flash (same alias, same family, litigation scores systematically shifted across a 2-day gap) before an $83 resume spend.

中文简介正在准备中。当前显示原文说明。

Raise real concurrency in asyncio LLM batch scorers built on the OpenAI SDK (AsyncOpenAI, including OpenAI-compatible providers like DeepSeek). Use when: (1) raising an asyncio.Semaphore above ~100 produces no throughput gain, (2) a batch pipeline saturates near 100 in-flight requests despite a larger semaphore, (3) planning a high-concurrency campaign against a provider with no hard rate limit (DeepSeek v4-flash tolerates 2000+ in flight). Root cause: AsyncOpenAI's default httpx pool caps max_connections at 100, silently bottlenecking any larger semaphore — you must pass a custom http_client with httpx.Limits sized to the semaphore.

中文简介正在准备中。当前显示原文说明。

为 OpenClaw 配置硅基流动 (SiliconFlow) 作为模型源。SiliconFlow 是国内领先的 AI 模型推理平台，提供 98+ 个 chat 模型，包含多个免费模型（Qwen3-8B、DeepSeek-R1-8B 等）。使用标准 OpenAI 协议（openai-completions）。包含 provider 注册、模型定义、别名配置、fallback 链接入和验证的完整流程。当管理员说想"加硅基流动"、"配 SiliconFlow"、"接入 SF 模型"、"加 Kimi"、"加 Qwen3"、"加免费模型"、"接入 DeepSeek V3.2"时使用此 skill。

中文简介正在准备中。当前显示原文说明。

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.

中文简介正在准备中。当前显示原文说明。

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.

中文简介正在准备中。当前显示原文说明。

DeepSeek HarnessのWeb版・公式Desktop版の会話や設定を操作する。対象インスタンスを特定し、認証済みHTTP RPCまたはアプリ画面で読取・依頼・追記・中断を行う。既存チャットの操作をheadlessの別実行と取り違えない。

中文简介正在准备中。当前显示原文说明。

Configure AIsa as a first-class model provider for OpenClaw, enabling production access to major Chinese AI models (Qwen, DeepSeek, Kimi K2.5, Doubao) through official partnerships with Alibaba Cloud, BytePlus, and Moonshot. Use this skill when the user wants to set up Chinese AI models, configure AIsa API access, compare pricing between AIsa and other providers (OpenRouter, Bailian), switch between Qwen/DeepSeek/Kimi models, or troubleshoot AIsa provider configuration in OpenClaw. Also use when the user mentions AISA_API_KEY, asks about Chinese LLM pricing, Kimi K2.5 setup, or needs help with Qwen Key Account setup.

中文简介正在准备中。当前显示原文说明。

指导通过评价回答的强化学习对大语言模型进行后训练。结合Megatron-LM与SGLang，涵盖数据准备、并行执行及自定义生成流程。

• 开始语言模型的GRPO训练
• 并行运行回答生成与训练
• 训练使用工具的智能体

Guides large language model reinforcement learning with miles, covering environment setup, low-precision training, consistency between training and inference, and speculative decoding, where a smaller model proposes tokens for a larger model to verify.

中文简介正在准备中。当前显示英语简介。

• Configure reinforcement learning for large MoE models
• Investigate instability in FP8 training
• Set up speculative decoding for faster rollout generation

Guide reinforcement learning, training models through rewards, with slime. Prepare data, configure Megatron-LM training and SGLang response generation, and set up asynchronous runs or custom multi-turn workflows for supported language models.

中文简介正在准备中。当前显示英语简介。

• Preparing reward-based training for GLM or Qwen models
• Overlapping response generation and training
• Training agents that use tools across multiple turns

Configure large language model pretraining with torchtitan across multiple GPUs. Covers ways to split training work, launch jobs on computing clusters, enable Float8 numerical processing, and save or resume training through distributed checkpoints.

中文简介正在准备中。当前显示英语简介。

• Pretraining Llama 3.1 8B on eight GPUs
• Configuring multi-node training with SLURM
• Enabling Float8 training on H100 GPUs

Configure large language model training with NVIDIA Megatron-Core across multiple GPUs. This skill helps choose how to split computation, set training parameters, monitor GPU efficiency, and address memory limits or unstable training.

中文简介正在准备中。当前显示英语简介。

• Planning multi-GPU training for a LLaMA-style model
• Distributing MoE model experts across GPUs
• Tuning training throughput and GPU utilization

Guides large language model reinforcement learning with miles, including MoE models that route work to specialized components. Covers low-precision training, matching training and inference behavior, and accelerating training response generation with draft models.

中文简介正在准备中。当前显示英语简介。

• Setting up reinforcement learning for large MoE models
• Configuring FP8 or INT4 training
• Checking consistency between training and inference

Configure large language model training across multiple NVIDIA GPUs with Megatron-Core. The skill helps choose how to split models and workloads, set training parameters, monitor performance, and address memory shortages or low GPU utilization.

中文简介正在准备中。当前显示英语简介。

• Planning distributed training for a LLaMA-style model
• Distributing MoE expert networks across GPUs
• Tuning GPU utilization and training throughput

Set up reinforcement learning, training guided by rewards, for language models with slime. The skill covers Megatron-LM training, SGLang response generation, data preparation, asynchronous execution, and custom workflows for agents that use tools across multiple turns.

中文简介正在准备中。当前显示英语简介。

• Prepare GRPO training for GLM or Qwen3 models
• Overlap response generation and model training
• Train agents with multi-turn tool use

Configure large language model pretraining with torchtitan across multiple GPUs. Covers ways to split training work, launch jobs on computing clusters, enable Float8 numerical processing, and save or resume training through distributed checkpoints.

中文简介正在准备中。当前显示英语简介。

• Pretraining Llama 3.1 8B on eight GPUs
• Configuring multi-node training with SLURM
• Enabling Float8 training on H100 GPUs

Change the language model used by a Worker Agent through the agt command-line tool. The controller resolves model settings, regenerates configuration, and recreates the container to apply the change without a manual restart.

中文简介正在准备中。当前显示英语简介。

• Change a Worker's language model
• Apply registered model settings automatically
• Check whether a model change was accepted

Manage consumers, credentials, AI providers, and model routing in Higress AI Gateway through its Console API. The skill explains session authentication, access permissions, and how to update existing settings while preserving required version information.

中文简介正在准备中。当前显示英语简介。

• Create consumers and replace their authentication keys
• Route requests to providers by model name
• Grant workers access to AI routes