注册并分享邀请链接,可获得视频播放与邀请奖励。

与「Efficiency」相关的搜索结果

Efficiency 贴吧
一个关键词就是一个贴吧,路径全站唯一。
创建贴吧
用户
未找到
包含 Efficiency 的内容
ICYMI: our July AI recap ⬇️ 🌦️ Advanced weather forecasting with @NOAA and Google Cloud, using high-performance H4D virtual machines to run its atmospheric models, providing meteorologists with the AI capabilities needed to deliver life-saving early warnings. 💻 Introduced 3 new Gemini models — Gemini 3.6 Flash, 3.5 Flash-Lite, and 3.5 Flash Cyber — to deliver the efficiency, latency, and reliability to build AI agents at scale. 🇺🇸 Launched the Alliance for America’s Skilled Trades With BlackRock, Carhartt, and Ford to build a stronger pipeline for the skilled trades, scale evidence-based training approaches, and expand opportunities nationwide. ☁️ Rolled out AlphaEvolve, our Gemini-powered AI code-optimization agent, to help Cloud customers solve their hardest problems by automatically searching for better solutions and returning human-readable, optimized code. 🎵 Delivered significant advancements in musicality, lyrics, and vocal quality in Google Flow Music with Lyria 3.5, our newest music generation model. 📡 Launched three new FireSat satellites, expanding a global initiative led by the Earth Fire Alliance (EFA) with @GoogleResearch to help detect wildfires before they spread. 🧪 Established over 15 global bioresilience partnerships across governments, research groups, and biosecurity organizations to prevent model misuse, detect outbreaks rapidly, and deliver swift, coordinated responses in partnership with @IsomorphicLabs.
显示更多
AI模型评分都是被专项攻坚创造出来的,于是我对比了Fable5,Grok4.5, Kimi K3针对同一个交易系统审计结果进行了对比。 先说结论: Fable5:最适合作为系统级主审核模型 Kimi:最适合作为代码缺陷与一致性专项审核模型 Grok:最适合作为代码梳理和方案发散模型,不适合单独决定策略修改 最佳组合:Fable5全面审核+Grok 4.5代码梳理+K3代码审核 具体细节: 1. Fable5:系统级判断能力最强 Fable5 最大的优势不是代码读得比另外两个模型更多,而是它能把: 代码规则; sizing snapshot; intent ledger; 实际 block 统计; 当前资产 headroom; SELL/REDEEM 回流路径; 放进同一个因果框架。 它使用了几个非常关键的实盘指标: ADD 近 7 天约占新增资金 43%; 84% 资金已经部署; ETH、SOL、XRP headroom 为 0; 近 40 个周期中主要阻塞是:blocked_capital_efficiency=47 blocked_asset_cap=28 deployment cap=0 runway=0 这让它能够区分: “某个机制理论上可能限制资金” 和 “当前实盘真正正在限制资金的机制”。 最终它得出: ADD 对资金流向重要,但当前周转主因在回收端、资产 cap 和效率过滤,不在 ADD 准入本身。 这是三个模型中最接近生产系统审核要求的判断。 弱点 Fable5 仍有一些过度推断: 把 ADD 描述为让资金“锁得更久”,实际上 ADD 的剩余 TTE 通常比 ENTRY 短; 把超 cap 资产总持仓约 $382 说成可以“直接解锁 $382”,没有区分总持仓、超额部分和可成交部分; 把模型中的 redeem_lag_days=2 一度当作实际回款延迟; “$5 仓位几乎不受每美元每日利润门约束”的推理不正确,因为该指标已经按资金归一化; 2-lot 最低 ENTRY 建议可能系统性损失覆盖率。 因此,Fable5 的系统方向判断最好,但具体数字和金融指标仍需二次校验。 最适合的角色 PRIMARY_SYSTEM_REVIEWER LIVE_OPERATIONAL_DIAGNOSIS CHANGE_PRIORITY_DECISION CROSS_MODULE_ROOT_CAUSE_ANALYSIS 2. Kimi:代码缺陷侦测能力最强 Kimi 对代码结构的还原比较准确: 固定 ADD 次数和 interval 已退役; ADD 采用 target-gap 模型; ENTRY 60%,ADD 补到 100%; allocator 是最终数量权威; style 仅作诊断; 现金、集中度、shock、深度共同限制订单。 更重要的是,Kimi 找出了其他两个模型没有明确指出的具体问题: shared_deployable_pool() 读取 account_snap["capital"]["deployable_cash"] 但该字段可能没有实际写入 → 回退到 free_cash → 策略层与 allocator 层资金口径可能不一致 它还发现了: 合同写 debounce 60 秒,代码/配置为 30 秒; 注释周期 16 分钟,实际 loop 600 秒。 这些是典型的静态审核、字段追踪和合同一致性检查优势。 弱点 Kimi 在资本效率和交易语义上的推理弱于它的代码检查能力。 典型错误是: ADD 价格更高,所以边际 edge/day 必然更差。 这忽略了剩余持有时间也缩短。更高 ask 并不必然意味着更低 edge/day。 它还认为: 60/40 会让剩余资金长期闲置; 提高 entry share 会改善周转; CONFIRMATION_NO 应收紧; 增加单市场软 cap 会改善组合周转。 这些结论缺少真实候选竞争、实际 block attribution 和反事实分配数据支持。 最适合的角色 STATIC_CODE_AUDITOR SCHEMA_AND_FIELD_FLOW_CHECKER CONTRACT_IMPLEMENTATION_DIFF LOCALIZED_BUG_DISCOVERY Kimi 很适合回答: “代码是否存在字段没有写入、默认值回退、文档与实现不一致、某个 gate 实际是否生效?” 但不适合单独回答: “应该如何改变交易策略和资本分配?” 3. Grok:代码梳理最完整,但最容易过度设计 Grok 对整个 ADD 路径的整理最详尽: 各层准入条件; risk latch; REDUCE reentry cooldown; 价格带; fingerprint; emergency cap; market target; ENTRY/ADD gap; allocator 的现金、集中度、shock 和深度约束; ADD 与 ENTRY 的评分和 continuity; SELL/REDEEM 对现金回收的影响。 它对当前代码执行模型的概括非常清楚: 能不能加由 headroom 决定;加多少由 target gap 离散为 lot;ADD style 只是解释标签。 因此,在“快速理解一个陌生复杂系统”方面,Grok 表现很好。 弱点 Grok 最大的问题是: 从“发现一个可能的机制副作用”快速跳到“建议修改策略”。 它提出了大量未经实盘证明的改动: TIME_TOPUP 冷却; ADD 1.5 倍 edge/day 门槛; ask≥0.97 限制为 1 lot; 降低 peak target; 提高 entry share; 单次仅补部分 gap; 弱化 continuity; 降低 TTE confirmation 权重。 这些建议表面上都很合理,但存在三个问题: 没有先证明这些机制实际造成了损失; 没有量化被 ADD 挤出的 ENTRY 是否更优; 可能重新引入此前已经修复的低 ADD recall 和 leader fidelity 偏差。 Grok很擅长生成完整优化空间,但容易把: POSSIBLE SIDE EFFECT 升级成: CONFIRMED ROOT CAUSE 再进一步升级成: SHOULD CHANGE PRODUCTION LOGIC 这是生产交易系统审核中最危险的倾向。 最适合的角色 SYSTEM_MAPPING CODE_AND_CONFIG_EXPLANATION HYPOTHESIS_GENERATION DESIGN_OPTION_ENUMERATION 不适合作为唯一的: PRODUCTION_CHANGE_APPROVER ROOT_CAUSE_FINAL_AUTHORITY STRATEGY_SEMANTICS_GATEKEEPER 三个模型的典型思维模式 Grok 发现机制 → 推演可能副作用 → 生成多种优化 → 倾向建议修改 优点:覆盖广、思路多。 风险:过度设计、假设升级过快。 Kimi 追踪代码和字段 → 找实现不一致 → 找局部缺陷 → 尝试从缺陷推导策略改进 优点:代码问题定位强。 风险:局部正确不等于系统结论正确。 Fable5 理解代码 → 读取运行数据 → 找实际 binding constraint → 区分主因和次因 → 按实盘收益排序 优点:最接近生产运营思维。 风险:仍会在个别指标含义和金额口径上过度断言。
显示更多
At the South Africa Financial Services Industry Summit, Huawei Digital Finance BU CEO Jason Cao said competition in banking has shifted to token efficiency and real-world value creation, and trillion-parameter models like Kimi K3 prove open source is now mainstream in banking.
显示更多
To celebrate a week of efficiency and let you run 100'000 Luna threads this weekend... that's right... wait for it... I have reset usage limits for Codex and ChatGPT Work. Enjoy.
0
3K
24.9K
1.2K
转发到社区
MiniMax H3: Omni-Reference, Commercial-Grade Generation, Unbeatable Cost Efficiency, Open Weights
0
118
2.4K
311
转发到社区
MiniMax H3: Omni-Reference, Commercial-Grade Generation, Unbeatable Cost Efficiency, Open Weights Your creative destiny, on your terms. Now Live at & MiniMax API.
0
42
723
78
转发到社区
Over the past few months, the questions we’ve heard from institutional clients have increasingly focused on collateral efficiency, cross-asset risk, liquidity, and how tokenized equities can fit into existing portfolio strategies. So our institutional team has put our thinking into this rToken playbook for cross-asset capital management. For institutions exploring how crypto and tokenized equities can work together within one capital framework, this is for you 👇
显示更多
0
20
87
11
转发到社区
A new approach to acoustic simulation from PKU could support future work in embodied intelligence! The research team led by Li Sheng and Wang Guoping from the School of Computer Science, Peking University, has won the Best Paper Award at the prestigious CASAXR 2026 (39th International Conference on Computer Animation, Social Agents, and Extended Reality) held in Geneva, Switzerland. Targeting key bottlenecks in traditional acoustic simulation—including numerical errors and poor adaptability to complex dynamic boundaries—the team proposed the novel SonicRadiation framework. This ghost-cell-free hybrid boundary-grid coupling method achieves simultaneous improvements in simulation accuracy and efficiency. It lays a solid physical foundation for high-fidelity acoustic simulation, embodied intelligence training, and accurate sound-physics coupling modeling for world models, empowering AI agents with authentic acoustic perception and interactive capabilities in complex dynamic scenarios. #PKUResearch# #EmbodiedAI# @CampusBiotech
显示更多
Endeavor Health developed a first-of-its-kind digital pathology software stack on Google Cloud to digitize 1.5 million slides annually. The system uses BigQuery and generative AI to improve efficiency by 40% and help patients understand reports ↓
显示更多
American Express posted strong Q2 results, beating EPS estimates at $4.53 as revenue grew 10% YoY to $19.64B! CREDIT & OUTLOOK: Credit loss provisions dropped to $1.1B, prompting management to raise full-year 2026 revenue growth guidance to 10%. CAPITAL EFFICIENCY: Return on equity reached 34.4%. THE RATING: Strong profitability metrics balance against market valuation, keeping the Seeking Alpha Quant score at a neutral HOLD. Is American Express the premium financial play to hold through 2026? $AXP
显示更多