与「MINIMUS」相关的搜索结果

「入り口で八王子会と言えば入れます」というパワーワードのもと ローランドさんのアパレルオープンパーティーへ🌝 ありちゃん付き添いありがとう🥹🫶🏻 #八王子会# #MINIMUS#

建议收藏，用到时照着做。做完你能让 DeepSeek 自己决定“要不要调用你写的函数”，并把结果用人话讲回来。 AI 能查天气、查订单，是它自己去查的吗？DeepSeek 官方一句话拆台：模型本身不执行任何函数 这是 DeepSeek 官方 API 文档里 Tool Calls（工具调用）一页。先说结论： 官方原话是，函数要由你自己提供，模型本身不执行具体函数。它只负责说“我想调这个函数、参数是这些”，真正去跑的是你的代码。 先说清词。 工具调用（Tool Calls），是让模型能用外部工具的功能，比如查天气、查数据库。 函数（function），是你自己写好的一段代码，模型只会“点名”它。 tool_call_id（工具调用编号），是模型每次点名时附带的一个号码，你回传结果时要带上它，模型才知道你在回答哪一次。 strict 模式（严格模式，Beta 测试版），是让模型点名时严格按你定义的格式填参数。 一个真实例子：官方示例问“杭州天气怎么样”，模型返回的是 get_weather 这个函数和参数 location 等于杭州，你的代码去查到“24℃”，再交给模型，模型才说“杭州现在 24 度”。 先把话说全： 官方示例用的是非思考模式，请求里加 extra_body 把 thinking 的 type 设成 disabled。 官方写明：从 DeepSeek-V3.2 开始，思考模式下也能用工具调用，细节看官方 Thinking Mode 一页。 官方写明：Chat Completion 接口不支持在对话中间插入不是模型生成的工具调用，要插的话请改用 Anthropic API 或 Responses API。 strict 模式目前是 Beta，要用测试版地址。 一、先看懂整个流程 官方给的执行流程是四步：用户问杭州天气；模型返回要调用 get_weather，参数是杭州；你的代码真的去调用 get_weather，把结果交给模型；模型用自然语言回答“杭州现在 24 度”。 二、照官方示例做一遍 第一步：去 DeepSeek 开放平台拿一个 API key，电脑装好 Python 和 openai 库。 第二步：创建客户端，base_url 写 DeepSeek 官方接口地址（官方文档第一屏就有），api_key 填你的。 第三步：写 tools 清单。每个工具包含 type 等于 function，再写 name（函数名）、description（描述，告诉模型什么时候用）、parameters（参数）。官方示例的函数叫 get_weather，描述是“获取某地天气，用户须先提供地点”，参数 location 是字符串，并且在 required 里。 第四步：发第一次请求，model 官方示例写 deepseek-flash，把 messages 和 tools 一起传，并加上 thinking 关闭的设置。 第五步：从返回里取 tool_calls 的第一项，先把模型这条消息原样追加进 messages。 第六步：你的代码去执行函数，再追加一条 role 是 tool 的消息，带上刚才的 tool_call_id，content 写结果，官方示例写的是 24℃。 第七步：再发一次请求，这次模型就会用自然语言回答。 三、想要更稳：strict 严格模式 第八步：base_url 改成官方文档里写明的 beta 版接口地址（就是原地址后面加 /beta）。 第九步：每个 function 都加上 strict 等于 true。 第十步：服务器会检查你的 JSON Schema（结构说明），格式不合规或用了不支持的类型，会直接返回错误信息。 官方支持的类型是：object、string、number、integer、boolean、array、enum、anyOf，另外 $ref 和 $def 可以复用和做递归。 四、strict 的硬规矩 对象里所有属性都必须写进 required，并且 additionalProperties 必须设成 false。 字符串可以用 pattern（正则）和 format，官方列出的格式有 email、hostname、ipv4、ipv6、uuid，但不支持 minLength 和 maxLength。 数字支持 const、default、minimum、maximum、exclusiveMinimum、exclusiveMaximum、multipleOf。 数组不支持 minItems 和 maxItems。 五、看到什么算成功 第一次返回的是函数点名，不是答案；第二次返回的才是自然语言答案，内容里有你的函数结果。 六、卡住怎么办 strict 报错：先检查是不是漏写 required 或忘了 additionalProperties 设 false，再检查有没有用到 minLength、maxItems 这类官方不支持的写法。 结论：工具调用不是模型替你干活，而是模型“点名”、你的代码“动手”。想让点名更准，就用 strict。把这条复制链接发给他，下次他说 AI 查不到东西，先让他确认函数是不是自己写了。 你卡在第几步？评论区告诉我。同系列前几篇在我主页。 收藏版：5 步速查 一、tools 里写清函数名、描述、参数 二、第一次请求拿到 tool_calls 三、自己执行函数，回传 role 为 tool 的消息并带 tool_call_id 四、第二次请求拿到自然语言答案 五、想更稳就用 beta 地址加 strict 等于 true 常见坑 3 条 一、以为模型会自己执行函数，其实要你的代码去跑。 二、回传结果忘了带 tool_call_id。 三、strict 里漏了 required 或 additionalProperties 设成 false。 复制即用提示词：我想让模型调用一个叫 你的函数名 的函数，它需要这些参数：写参数。请帮我写出符合 DeepSeek Tool Calls 格式的 tools 清单，每个参数都写 description，并改成 strict 模式需要的写法。 S1哥观点：这篇最狠的地方，是官方把话挑明了：模型不执行函数，只负责点名。很多人以为 AI 在后台自己动手，其实动手的永远是你的代码。 真正的信号是：工具调用的难点不在模型，在你把函数描述写得多清楚。就像派活给新人，说明写得含糊，他再聪明也会点错人。 你现在想让 DeepSeek 调用的是什么，查数据还是发消息？评论区告诉我。 今天送：DeepSeek 工具调用速查卡，评论区第一条直接拿。 参考：DeepSeek 官方 API 文档 Tool Calls（S1哥 整理）

Binance has completed coverage of assets impacted in the security incident and we have published a new Proof of Reserves snapshot. Once the zero-knowledge proof is generated, users can independently verify that their assets are backed at a minimum of 1:1 - To our users and community: thank you for your patience and understanding.

Trade your shielded $ZEC for confidential USDC on Ethereum, without ever unshielding it. No bridge. No custodian. No wrapped ZEC. And the amount never shows up on chain. I connected @zama and @Zcash to make it work. Live on testnet 👇 What it is ZEC Desk is a small trading desk on Zama's FHEVM, on Ethereum Sepolia. On one side there's cUSDC, a dollar token whose balances are encrypted. On the other side there's shielded ZEC on Zcash testnet. You can trade either way. Built on Zama's model Zama's Confidential RFQ is already live on Ethereum mainnet, open to everyone: It runs confidential swaps with encrypted quotes, a sealed-bid pick on chain and a hidden floor. ZEC Desk takes that same model and extends it to native shielded ZEC. Zama's swap settles both legs on one chain, inside its own contract, so a Zcash leg can't plug into it directly. Here that leg lives on Zcash and settles minutes later with a proof, so the mechanics run in my own contract. It's a sketch of how native ZEC could sit next to Zama's swaps one day, with no wrapped token. How a trade works (buying ZEC) 👇 1️⃣ You lock cUSDC in the desk. The amount is encrypted. 2️⃣ Makers send encrypted quotes. Nobody sees anyone else's price, or the minimum you'll accept. 3️⃣ The contract picks the best one under FHE. 4️⃣ The winning maker sends shielded ZEC to your Zcash address, with the order's reference in the memo. 5️⃣ A zero-knowledge proof shows that payment happened, and the contract releases your cUSDC to the maker. If no proof arrives in time, your cUSDC comes back to you. Selling is the same in reverse: the maker locks cUSDC, you pay the ZEC, and the proof releases the cUSDC to you. Why these pieces Each piece does one job. Zcash keeps the payment private. Zama keeps the amounts encrypted and can still compute on them. Seam, the Zcash light client I built last month, connects the two: a contract checks every Zcash block header itself, and a RISC Zero proof shows a shielded payment reached the right receiver. The transaction, the note and the address never touch the EVM chain. The masked amount This was the hard part. Seam's proof used to write the paid amount in the clear, and ZEC amount × price gives away the dollar amount. So the amount is masked. The contract adds an encrypted one-time mask to the encrypted amount and reveals only the sum, which looks random to everyone else. The mask comes from a secret that the side receiving the ZEC hands to the payer. The proof checks that the payment was at least the sum minus the mask, and never states the amount. One-time pads are old, and Zama's RFQ uses one too. Using one to hand a value from FHE to a ZK proof is the part I haven't seen before. If you have, please send it my way. No vault, no signers Seam can read Zcash, but it can't take ZEC from anyone. So the design never needs someone to pay ZEC on someone else's behalf. Whoever pays, pays from their own wallet, and the escrow on the EVM side enforces the deal. The side that paid the ZEC can prove it itself, so nobody depends on the other side's goodwill. Seam does more than this Any EVM app that needs to know a shielded payment happened can use Seam: pay-with-ZEC checkouts, escrows, payouts, cross-chain orders. This desk is one example. Where it stands ↓ Testnet only, unaudited. On 12 November 2026 I ran a Buy end to end: ➡️ Settled with a proof: ➡️ One proof: 2 minutes on a rented RTX 4090, about $0.25 for the whole session ➡️ Gas per Buy on Sepolia: about 7.6M, in six transactions ➡️ A Sell also ran on Sepolia, up to the maker locking its cUSDC; it settles through the same proof step ➡️ Seam's light client is now deployed on Ethereum Sepolia too, next to Base Sepolia Where it can go ↓ → The desk only needs Zama's FHEVM and a cUSDC. Both already run on Polygon, where following Zcash costs a fraction of Ethereum mainnet. → Zama keeps adding chains. On any cheap EVM chain it reaches, the same desk can run, and if it comes to Base, Seam's relay already runs on Base Sepolia. → Zama's roadmap also goes beyond the EVM, to chains like Solana. Seam's verifier would need a port there. → On Ethereum mainnet, following Zcash header by header would cost over 2B gas a day, so batched header proofs are next for Seam. What it can't do yet - Settlement takes minutes, not seconds. Zcash confirmations set the pace, not the GPU. - Right now I'm the only maker, and a maker needs ZEC, cUSDC and a GPU. - Amounts are hidden, but EVM addresses are public. Read more ↓ Seam on Seam: ZEC Desk:

A $1.2 million sale. At 2.5% + GST: $33,000 Agent Up at 1% + GST: $13,200 Difference: $19,800 Full-service real estate. Different business model. 1% + GST. Minimum $11,000 inc GST.

ClusterMAX 3.0 is here! ClusterMAX 3.0 debuts with a comprehensive review of the neocloud industry, covering 77 providers. We increase our market view to cover 323 providers, up from 209 in ClusterMAX 2.0, 169 in ClusterMAX 1.0, and 124 in the original AI Neocloud Playbook and Anatomy article. We have now interviewed well over 200 end users of neoclouds as part of this research. We update our itemized list of criteria across 10 categories, and update our direct descriptions of our expectations for Slurm, Kubernetes, Standalone Machines, Monitoring Dashboards, and Health Checks. All of this content is live on our website. We encourage providers to use these lists when developing their offerings. We still consider these lists as an amalgamation of our experience interviewing end users, making them representative of the features that end users expect from their cloud providers. Nebius joins CoreWeave in the Platinum tier. While CoreWeave still sets the technical bar for others to follow, Nebius is now established as a provider that consistently commands a premium pricing over others. Strong business decisions by Nebius have put them in a position to serve an entire class of neolabs at seller’s prices. Google Cloud joins Oracle in the Gold tier. Azure moves to Silver, Fluidstack moves to Unavailable, and Crusoe drops to Bronze. Lambda, Firmus and TensorWave remain in Silver, while GMI moves up to Silver from Bronze. Many companies drop from Silver (or Gold) to Bronze or lower. We raise the bar this round as only 19 neoclouds globally achieve a Medallion rating. We establish a tier between Bronze and Underperforming: the Participation Ribbon tier. 15 providers join this rating, which more accurately describes our opinion that they do the bare minimum to get by.

here's a prompt to improve your agent harness based on what we've learned at cursor. enjoy # Improve this agent harness's token efficiency You're working on an LLM agent harness: the system prompt, tool definitions, request assembly, context caching, compaction, and retrieval, and how work is split across agents. Make the agent's runs cheaper without making it worse at its job. - Objective: lower price-weighted token cost per completed task. - Constraint: no measurable drop in task quality. Measure per task, not per request. Every turn resends the prefix (tools, instructions, setup, and the conversation so far), so a change that shrinks each request but adds turns can cost more. Weight tokens by billing type: output, uncached input, and cached input are priced very differently. Work in this order: map the harness and measure the baseline, rank the opportunities, make the changes that are safe to make directly, put the rest behind flags or in proposals, then report. Figures below come from one team's production coding agent and its multi-agent experiments. Use them to gauge magnitude, not as targets. One round of these changes (prompt trimming, tool offloading, cache layout, sparse line numbers, subagent tuning) cut that team's overall token cost about 7% with no loss in quality. The larger percentages apply only to the part of the request each change touched. ## Principles 1. Change what the harness sends, not how hard the model tries. Don't ask the model to conserve tokens. A harness that told its model to "take care to preserve tokens and not be wasteful" found it grew reluctant to take on ambitious tasks and sometimes quit, saying it wasn't supposed to waste tokens. 2. Capable models need definitions, not commands. Lists of "DO NOT", "You must", and "Important", and guards against older models' habits, can usually be replaced with plain descriptions of what each tool does. One team cut about two-thirds of its system prompt this way, and the shorter prompt worked across model families. Instruct only on what the model can't know (the product, the environment, the user's processes) and on quirks you've seen in transcripts. 3. Static context is for what most turns need. Everything else should be discoverable when needed. Less up-front context also means less confusing or contradictory information. 4. Expect removals to win. Guardrails written for weaker models, coordination steps that became bottlenecks, and prompting for behavior the model now does on its own all cost tokens. 5. Real usage decides. Evals are a fast proxy, but they skew toward hard problems and miss the real mix of requests. ## 1. Map the harness and measure the baseline Find: - Where requests are assembled, the system prompt, and tool schemas. If a framework or SDK builds requests, find its hooks for message order, cache control, and tool loading. - How tool results are formatted, and how history is kept, trimmed, or summarized. - How subagents or parallel agents are spawned, if any. - Which models and provider APIs are used. From the provider's docs, get the prompt caching behavior (automatic or explicit breakpoints, TTL, minimum cacheable length) and the prices for output, uncached input, and cached input. - Existing logging, token accounting, and evals. If the harness doesn't record per-request token usage by billing type and cache hits, add that first. Everything later depends on it. Then render a few real requests (from logs, or by running representative tasks) and count tokens per section with the model's tokenizer or the API's usage fields. Produce: - Cost share by source × billing type. Sources: system prompt, tool definitions, skill/rule/integration descriptions, user messages, file reads, search results, command and other tool output, history, summaries, subagents. - Static tokens per request, cache hit rate, and turns per task. - Per tool: the share of runs that call it at least once, and its error rate. Read the rendered requests, not just the templates. Duplication, leaked volatile values, and misordered blocks only show up there. Rank opportunities by share of spend × fraction removable ÷ quality risk. ## 2. System prompt and injected context Label every instruction: - Keep: product or environment knowledge the model can't infer, fixes for quirks seen in this model's transcripts, and rules a mode depends on. - Rewrite: commands and emphasis into plain descriptions. Reminders into constraints: "No TODOs, no partial implementations" works better than "remember to finish implementations." Vague quantities into ranges: "generate 20–100 tasks" gets far more ambitious behavior than "generate many tasks." - Delete: things capable models do by default, guards against behavior you haven't seen from this model, text that repeats tool descriptions, and lines that could contradict a user request. Models trained to rank system instructions above user messages will side with the system prompt. - Move: anything per-user or per-request (date, environment, repo state, lists of skills or subagents, user rules) into a user-role setup message after the cache boundary. Audit other injected context the same way. As models improved, the team behind these figures dropped directory trees, pre-retrieved snippets, compressed copies of attached files, lint errors injected after every edit, forced expansion of short file reads, and caps on tool calls per turn. They kept small, high-value facts: OS, repo status, and open or recently viewed files. Skip checklists for open-ended work. The model optimizes the listed items and deprioritizes everything else. ## 3. Tool definitions Tool schemas ride along on every request. Most tools beyond the core set were each needed in under 20% of conversations, and moving them out of static context cut tool-description tokens 60%. Doing the same for integration tools (such as MCP servers), with names in context and full schemas in one folder per server that the agent can search with grep or jq, cut total tokens 46.9% in sessions that used them. - Keep in static context: high-frequency tools (for a coding agent: read, search, edit, shell), tools the model tries to call even when they're absent, and tools a mode depends on. - Offload the rest: leave a name or one-line pointer and make the full schema discoverable on demand. Group related tools so they load together, and put status (such as "needs re-authentication") where the agent will see it. - Tighten what remains: describe behavior and arguments, and drop usage lectures. - Pick the split by testing a few configurations and tracking tokens, cost, latency, tool-call errors, and task success. ## 4. Cache layout Order each request so the reusable prefix is as long as possible: `tool definitions → system instructions → [breakpoint] → setup message (skills, subagents, rules, environment) → [breakpoint] → conversation` - Keep the prefix byte-identical across turns. Use deterministic tool order and serialization, put timestamps and IDs after the boundary, and don't rewrite earlier messages except when compacting. - Use explicit breakpoints if the provider supports them. Otherwise rely on automatic prefix caching with the stable part first. Respect TTL and minimum-length rules. - Switching models mid-conversation throws away the cache (caches are per model and provider) and hands the new model a history it didn't write. When a different model is needed, run it as a subagent with fresh context. Explicit breakpoints plus moving per-request setup after them cut cold cache misses 20%. ## 5. Tool results and other context added during a run - Large outputs (commands, integrations, logs): write them to a file and return the path, size, and a short tail. The agent can tail, grep, or read ranges for more. Truncating loses data, and inlining bloats every later request. Treat long-running terminal sessions the same way. - High-volume formats: look for overhead repeated on every line or item. Numbering every 10th line of a file read instead of every line cut cache-read tokens 1.6% without hurting citation accuracy. Each number costs 3–5 tokens, and agents read tens of thousands of lines per session. Also check repeated absolute paths, verbose JSON keys, ANSI codes, progress bars, and repeated headers. - Good retrieval saves exploration turns. Adding semantic search alongside grep raised codebase question-answering accuracy 12.5% on average and cut the iterations users needed. - Tool errors waste tokens and leave confusing debris in context. Classify expected errors (invalid arguments, unexpected environment, provider error, timeout, user abort), treat unknown errors as harness bugs, and track rates per tool and per model. One focused effort along these lines cut unexpected tool errors 10×. ## 6. Long runs: compaction, subagents, and model mix - Compaction: keep the summarization prompt short and the summary compact, carry forward plan state and remaining tasks, and save the full history to a file the agent can search for details the summary dropped. A model trained to self-summarize from a one-line prompt wrote ~1k-token summaries with half the compaction error of a multi-thousand-token prompt that produced 5k+ token summaries. Untrained models may need more guidance, so test how short you can go. A more expensive summarization model made a negligible difference. - Scratchpads and running notes: rewrite them instead of appending. For repeated work in one environment, a small agent-maintained notes file with a line budget, loaded at start, is a promising way to shorten later runs. - Subagents: fresh context keeps the parent lean, but isolation adds coordination cost (duplicate or stale work). If the model already delegates on its own, remove prompting that pushes it to. Have subagents return short handoffs: what was done, findings, concerns, and deviations. A subagent should use a different model only when the user or harness says so. - Model mix: in large multi-agent runs, workers used at least 69% of tokens, and over 90% in most runs. A frontier planner with cheap workers matched a frontier model doing everything at about one-eighth the cost. Planner choice still changes worker spend. One planner that cost less on its own saw its workers use several times more tokens, and the run cost more overall. Measure the whole tree. - Routing and reasoning effort: send simple turns to a cheaper model or lower effort, and upgrade only when a stronger model is clearly better. A router built this way matched or beat single frontier models on user satisfaction at 41–68% lower cost. - Reasoning continuity: if the API returns reasoning items (including encrypted ones), pass them back on later turns and alert when they go missing. Dropping them cost one reasoning model 30% on a coding benchmark, and it burned tokens reconstructing its plan. ## 7. Fit the harness to each model Adapt to what each model was trained on instead of forcing one shape on all of them. If you've tuned the harness for a similar model, start from that version. - Edit format: use the one the model was trained on (for example, patch-style or search-and-replace). An unfamiliar format costs extra reasoning tokens and causes more mistakes. - Shell or tools: shell-first models fall back to `cat` or inline scripts. Name tools after their shell equivalents (such as `rg`), and if needed add: "If a tool exists for an action, prefer to use the tool instead of shell commands (e.g. read_file over `cat`)." - Literalness: some model families follow instructions literally and others tolerate imprecision. Some spiral on emphasized wording. Strip caps and emphasis for literal models. - Triggers: some models ignore a tool until told when to use it. A literal trigger works: "After substantive edits, use the to check recently edited files for linter errors. If you've introduced any, fix them if you can easily figure out how." - Progress updates: if a model reports progress through reasoning summaries, keep them to 1–2 sentences that note new findings or a change of tactic, and remove instructions about messaging mid-turn. - Quirks worth a targeted line: hedging or refusing as context fills ("context anxiety"), declaring completion early, stopping to ask permission, and calling tools that don't exist. Tie each added instruction to the transcript behavior it fixes. Re-audit when models change, since guidance one version needed can be dead weight for the next. ## 8. Validate - Offline: run a fixed set of realistic tasks before and after, ideally drawn from real usage and phrased the way users actually write (short and ambiguous). Compare task success, tokens, cost per task, turns, and tool errors. Don't ship a change that lowers success. - Online, if you have users: A/B test each change or small bundle. The primary metric is cost per completed task. Guardrails are task success signals, tool-call errors, latency, turns per task, and cache hit rate. For a coding agent, a good success signal is how much agent-written code survives over time. In general, check whether the user's next message moves on or reports a problem. - Ship only when cost drops and no guardrail regresses beyond noise. Record null results. ## What to change directly and what to propose - Change directly, each in its own revertible commit: token and cache telemetry, deterministic serialization and tool order, moving volatile content out of the cached prefix, explicit cache breakpoints, writing large outputs to files instead of truncating, passing back reasoning items that are being dropped, and fixes for recurring tool errors. - Change behind a flag so it can be tested: system prompt edits, tool offloading, output format changes, compaction changes, and subagent prompting. - Propose only: changes to which models run, routing, reasoning-effort defaults, or how work is split across agents. ## Traps - Asking the model to use fewer tokens or do less. - Truncating tool output. - Dropping reasoning items to save input tokens. - Volatile content in the cached prefix, or tool order that changes between requests. - Offloading a tool the model needs on the first turn or tries to call when it's missing. - Emphasis-heavy prompts (MUST, NEVER, IMPORTANT, all caps), especially with literal models. - Forcing a terser output format than the model was trained on. Fewer output tokens can mean less thinking and worse results. - Optimizing raw token counts instead of cost, per request instead of per task, or evals instead of real usage. - Switching models mid-conversation to save money. - Adding coordination layers that become bottlenecks. ## Report back with 1. The harness map and baseline: cost by source × billing type, with the biggest sources called out. 2. A ranked list of changes: layer, what changes, estimated savings and how you estimated them, quality risk, how to validate, and how to roll back. 3. The changes you made, including a system prompt diff with a keep, rewrite, delete, or move reason for each line. 4. A test plan for the flagged changes. 5. Gaps: anything you couldn't find or measure.

The Financial Times reports that technology groups have offered up to $300 billion of guarantees in less than a year to support debt financing for AI data centres and chips. Structures include minimum residual-value guarantees and special-purpose vehicles; rating agencies can adjust leverage when stressed collateral values fall below guaranteed amounts. There can be separate RVGs in the same deal - powered shell layer, chips layer, lease layer. Total off-balance sheet commitments top $3tn across the largest tech companies per MS and the FT