{"@context":"https://schema.org","@type":"NewsArticle","generatedAt":"2026-09-21T18:01:22.787Z","headline":"普林斯顿、蚂蚁集团与斯坦福团队提出 AQuA：面向量化金融的两段式智能体自动因子发现框架","description":"普林斯顿大学、蚂蚁集团和斯坦福大学的研究团队提出 AQuA，一套由两个互不共享智能体、记忆和状态的语言模型量化研究系统，其设计要点是在迭代开始前冻结数据划分、特征与标签定义和评估器，让智能体只能在受约束的 DSL 内探索，以阻断自我改进中的数据泄漏。","url":"https://www.aioga.com/news/cmtiuswre05o4ro9yr14b8s5f/","mainEntityOfPage":"https://www.aioga.com/news/cmtiuswre05o4ro9yr14b8s5f/","datePublished":"2026-09-01T15:54:14.000Z","dateModified":"2026-09-01T15:54:14.000Z","inLanguage":"zh-CN","publisher":{"@type":"NewsMediaOrganization","name":"Aioga","url":"https://www.aioga.com"},"citation":["https://www.marktechpost.com/2026/09/01/aqua-a-two-part-agentic-framework-for-autonomous-factor-discovery","https://aihot.virxact.com/items/cmtiuswre05o4ro9yr14b8s5f"],"canonicalUrl":"https://www.aioga.com/news/cmtiuswre05o4ro9yr14b8s5f/","directAnswer":{"@type":"Answer","text":"普林斯顿大学、蚂蚁集团与斯坦福大学团队提出 AQuA，由两个不共享智能体、记忆、候选空间和研究状态的语言模型量化研究系统组成，分别探索加密资产因子与美股时间序列模型。","url":"https://www.aioga.com/news/cmtiuswre05o4ro9yr14b8s5f/","dateCreated":"2026-09-01T15:54:14.000Z","author":{"@type":"Organization","@id":"https://www.aioga.com/authors/aioga-editorial/#editorial-team","name":"Aioga Editorial Team","url":"https://www.aioga.com/authors/aioga-editorial/"}},"evidence":[{"@type":"CreativeWork","name":"marktechpost.com source article","url":"https://www.marktechpost.com/2026/09/01/aqua-a-two-part-agentic-framework-for-autonomous-factor-discovery","datePublished":"2026-09-01T15:54:14.000Z","provider":{"@type":"Organization","name":"marktechpost.com","url":"https://www.marktechpost.com/2026/09/01/aqua-a-two-part-agentic-framework-for-autonomous-factor-discovery"}},{"@type":"CreativeWork","name":"AIHot archive record","url":"https://aihot.virxact.com/items/cmtiuswre05o4ro9yr14b8s5f","datePublished":"2026-09-01T15:54:14.000Z","provider":{"@type":"Organization","name":"AIHot","url":"https://aihot.virxact.com/items/cmtiuswre05o4ro9yr14b8s5f"}}],"aggregationSource":"MarkTechPost（RSS）","originalPublisher":{"name":"marktechpost.com","url":"https://www.marktechpost.com/2026/09/01/aqua-a-two-part-agentic-framework-for-autonomous-factor-discovery"},"geoDeepAnswer":null,"article":{"id":"cmtiuswre05o4ro9yr14b8s5f","slug":"cmtiuswre05o4ro9yr14b8s5f","url":"https://www.aioga.com/news/cmtiuswre05o4ro9yr14b8s5f/","title":"普林斯顿、蚂蚁集团与斯坦福团队提出 AQuA：面向量化金融的两段式智能体自动因子发现框架","title_en":"","summary":"普林斯顿大学、蚂蚁集团和斯坦福大学的研究团队提出 AQuA，一套由两个互不共享智能体、记忆和状态的语言模型量化研究系统，其设计要点是在迭代开始前冻结数据划分、特征与标签定义和评估器，让智能体只能在受约束的 DSL 内探索，以阻断自我改进中的数据泄漏。","source":"MarkTechPost（RSS）","sourceUrl":"https://www.marktechpost.com/2026/09/01/aqua-a-two-part-agentic-framework-for-autonomous-factor-discovery","aiHotUrl":"https://aihot.virxact.com/items/cmtiuswre05o4ro9yr14b8s5f","publishedAt":"2026-09-01T15:54:14.000Z","category":"行业动态","score":58,"selected":false,"articleBody":["Quantitative research agents that write their own experiments can corrupt the evidence they later learn from. A leaky feature that scores well gets stored as a successful precedent and propagated through later iterations. Prompt-level instructions and reviewer agents do not close this, because author and reviewer share the same blind spots. A team of researchers from Princeton University, Ant Group and Stanford University propose AQuA ：https://arxiv.org/abs/2608.12841 . AQuA is a pair of language-model-driven research systems that improve their own research process across iterations while the thing judging them stays frozen. One discovers symbolic alpha factors on crypto; the other develops time-series models on US equities. They share no agents, memories, candidate spaces or research state.","Quantitative research breaks on small methodological errors that produce convincing but non-reproducible backtests, documented since Bailey et al.：https://www.ams.org/journals/notices/201405/rnoti-p458.pdf. An agent writing its own experiments makes this worse: a leaky feature that scores well gets stored as precedent, and recursion amplifies an undetected bug as readily as a real discovery.","Prompt-level instructions and model review are not an integrity boundary. Repeated access to a fixed holdout causes adaptive overfitting：https://proceedings.mlr.press/v37/blum15.html, and LLM agents have been observed exploiting misspecified objectives and evaluators：https://arxiv.org/abs/2503.11926. AQuA instead makes leakage-inducing actions unavailable. Each part fixes its splits, feature and label definitions and evaluator before any iteration starts, and the agent emits only a constrained factor expression or a single config diff. The research team call this asymmetric freedom : the agent explores freely inside its DSL, but the evaluator sits outside the adaptive surface. What improves is the research process.","Part I is a six-agent pipeline: Data Steward, Visual Analyst, Idea Miner, Factor Evaluator, Backtest Engineer and Research Librarian — orchestrated by an AI Manager. Agents never call one another; every handoff goes through the Manager, keeping runs auditable.","A factor enters as a falsifiable proposal, not an expression: hypothesis, mechanism, predicted direction, and refutation conditions. Only then is it assembled from the standard formulaic-alpha operator registry：https://arxiv.org/abs/1601.00991. Because every time-series operator reads only a trailing window and every cross-sectional operator reads only the current timestamp, causality is closed under composition. Three feedback loops run: direction calibration inside a backtest, falsification-driven belief update inside a run, and cross-run memory that steers the next search.","On a crypto five-minute universe the combined validation Spearman IC climbs across 20 research epochs to approximately 0.190 , against 0.171 for an adapted AlphaMemo：https://arxiv.org/abs/2606.20625, 0.151 for an adapted AlphaGen：https://arxiv.org/abs/2306.12964, 0.137 for LSTM, 0.106 for LightGBM and 0.075 for an Alpha158-style baseline. Individual mechanisms stay weak — single-factor ICs of 0.026 to 0.037. The claim is about the harness, not one expression.","Part II predicts each stock’s forward return over the next thirty minutes on intraday US equities. Training runs on 2010–2019, 2020 is an embargo gap nothing touches, and 2021–2025 is untouched test data. Selection uses an inner-validation slice from the end of the training window only.","A hypothesis here is one config diff — architecture, loss, sampler or optimizer — and one diff produces exactly one variant, keeping variants comparable. The predictor is a hybrid: a multi-scale 1-D convolutional front-end, a configurable backbone spanning LSTM, Mamba：https://arxiv.org/abs/2312.00752 and attention：https://arxiv.org/abs/1706.03762 (attention in the reported run), a cross-sectional stage that mixes across the panel, gated fusion and a pooled per-stock readout.","No single price-volume feature carries the signal: the strongest is a 5-minute return at −0.031, and a ridge combination reaches only +0.025. Across model families on identical data and the same evaluator, per-stock raw IC runs +0.0251 (ridge), +0.0397 (LGB), +0.0434 (xLSTM：https://arxiv.org/abs/2405.04517), +0.0535 (LSTM), +0.0613 (GRU) and +0.0843 for the hybrid — +0.0230 absolute over the best baseline, 37.5% relative. The two parts’ ICs use different conventions and the paper states plainly they should not be compared.","The per-stock score becomes a dollar-neutral threshold long/short book at a two-leg cost of 2 bps. Sector-neutralizing raises the held-out Sharpe to +2.15 , with training and held-out values nearly equal. A causal volatility-targeting overlay lifts it to +2.50 , and a fully causal walk-forward choosing every parameter from past data alone still reaches +2.00 . Per-stock R² is 1.20%. Sharpe by year runs +1.7, +3.5, +1.9, +1.8 and +2.7 for 2021 through 2025 — positive in every year, including the 2022 drawdown.","Need to partner with us for promoting your GitHub Repo OR Hugging Face Page OR Product Release OR Webinar etc.? Connect with us ：https://forms.gle/wbash1wF6efRj8G58","Note:Thanks to the Ant Research team for the thought leadership/ Resources for this article. Ant Research team has supported this content/article for promotion.","Asif Razzaq is the CEO of Marktechpost Media Inc.. As a visionary entrepreneur and engineer, Asif is committed to harnessing the potential of Artificial Intelligence for social good. His most recent endeavor is the launch of an Artificial Intelligence Media Platform, Marktechpost, which stands out for its in-depth coverage of machine learning and deep learning news that is both technically sound and easily understandable by a wide audience. The platform boasts of over 2 million monthly views, illustrating its popularity among audiences."],"articleImages":[{"sourceUrl":"https://www.marktechpost.com/wp-content/uploads/2019/06/Screen-Shot-2021-09-14-at-9.02.24-AM-300x300.png","alt":"","afterParagraph":11,"url":"/media/articles/cmtiuswre05o4ro9yr14b8s5f/787a6d54564e8e19.webp"},{"sourceUrl":"https://www.marktechpost.com/wp-content/uploads/2026/08/blog2211-3-100x70.png","alt":"Keenable AI Open-Sources NEEDLE: A Live Search Benchmark That Rebuilds Its Query Set Every Hour","afterParagraph":12,"url":"/media/articles/cmtiuswre05o4ro9yr14b8s5f/29f802a690594064.png"}],"mediaStatus":"ok","articleBodyZh":["定量研究代理如果自己编写实验，可能会破坏它们随后从中学习的证据。一个表现良好的泄露特征会被存储为成功的先例，并在后续迭代中传播。提示级指令和评审代理无法解决这一问题，因为作者和评审者具有相同的盲点。来自普林斯顿大学、蚂蚁集团和斯坦福大学的一个研究团队提出了AQuA：https://arxiv.org/abs/2608.12841。AQuA是一对由语言模型驱动的研究系统，它们在迭代中改进自己的研究过程，而评估它们的系统保持不变。一个发现加密货币的符号alpha因子；另一个开发美国股票的时间序列模型。它们不共享代理、记忆、候选空间或研究状态。","定量研究在小的技术错误上会崩溃，这些错误会生成令人信服但不可重复的回测，Bailey等人已有记录：https://www.ams.org/journals/notices/201405/rnoti-p458.pdf。一个自己编写实验的代理会使情况更糟：表现良好的泄露特征会被存储为先例，而递归会像真实发现一样放大未检测到的错误。","提示级指令和模型评审不是完整的诚信边界。反复访问固定的保留集会引起自适应过拟合：https://proceedings.mlr.press/v37/blum15.html，并且已观察到大语言模型代理利用错误设定的目标和评估器：https://arxiv.org/abs/2503.11926。AQuA则使引发泄露的行为不可用。每个部分在任何迭代开始前都会固定其数据分割、特征和标签定义以及评估器，代理仅输出受约束的因子表达或单一配置差异。研究团队称之为不对称自由：代理在其DSL内部自由探索，而评估器则位于自适应表面之外。改进的对象是研究过程。","第一部分是一个六代理管道：数据管理员、可视化分析师、想法挖掘者、因子评估员、回测工程师和研究图书管理员——由AI经理协调。代理之间从不直接调用；每次交接都通过经理进行，使运行过程可审计。","因素作为可证伪的提议进入，而非表达式：假设、机制、预测方向和反驳条件。只有在这时，才从标准的公式-α算子注册表中组装出来：https：//arxiv.org/abs/1601.00991。由于每个时间序列算子只读取尾随窗口，每个截面算符只读取当前时间戳，因果关系在合成下是封闭的。运行三个反馈循环：回测中的方向校准，运行中的反证驱动信念更新，以及引导下一次搜索的交叉运行记忆。","在加密五分钟宇宙中，Spearman IC的综合验证跨越20个研究纪元，达到约0.190，而改良版AlphaMemo为0.171：https：//arxiv.org/abs/2606.20625，改良版AlphaGen为0.151：https：//arxiv.org/abs/2306.12964，LSTM为0.137，LightGBM为0.106，Alpha158风格基线为0.075。各机制保持弱势——单因子IC为0.026至0.037。该声明关注的是线束，而非单一表达式。","第二部分预测每支股票在未来三十分钟内对美国盘中股票的远期回报。2010–2019年的培训运行，2020年为禁运缺口，未被触及，2021–2025年为未动测试数据。选择仅使用培训窗口末的内部验证切片。","这里的假设是一个配置差异——架构、损耗、采样器或优化器——而一个配置只产生一个变体，保持变体的可比较性。预测变量是混合型的：多尺度一维卷积前端，可配置的跨LSTM主干网，Mamba：https：//arxiv.org/abs/2312.00752，注意点：https：//arxiv.org/abs/1706.03762（报告运行中注意），一个跨面板混合的横截面级，门控融合和按库存数据汇集。","没有单一的价格-交易量特征携带信号：最强的是5分钟收益为−0.031，而岭回归组合仅达到+0.025。在相同数据和相同评估器下，不同模型家族的每股原始IC分别为+0.0251（岭回归）、+0.0397（LGB）、+0.0434（xLSTM：https://arxiv.org/abs/2405.04517）、+0.0535（LSTM）、+0.0613（GRU），混合模型为+0.0843——比最佳基线绝对提高+0.0230，相对提高37.5%。两部分的IC使用不同的标准，论文明确表示不应进行比较。","每股评分变为一个美元中性阈值的多空组合，双向成本为2个基点。进行行业中性化将持出Sharpe提高到+2.15，训练值和持出值几乎相等。因果波动率目标覆盖可将其提高至+2.50，而完全因果的滚动前瞻模型（只使用过去数据选择每个参数）仍可达到+2.00。每股R²为1.20%。按年份计算的Sharpe分别为2021至2025年的+1.7、+3.5、+1.9、+1.8和+2.7——每年均为正值，包括2022年的下滑期。","需要与我们合作推广您的GitHub仓库、Hugging Face页面、产品发布或网络研讨会等吗？请联系：https://forms.gle/wbash1wF6efRj8G58","注意：感谢蚂蚁研究团队对本文提供的思想领导力/资源支持。蚂蚁研究团队为本文/内容的推广提供了支持。","Asif Razzaq是Marktechpost Media Inc.的首席执行官。作为一名有远见的企业家和工程师，Asif致力于利用人工智能的潜力造福社会。他最近的项目是推出人工智能媒体平台Marktechpost，该平台因其对机器学习和深度学习新闻的深入报道而脱颖而出，内容既具有技术深度又易于广大受众理解。该平台每月访问量超过200万次，显示出其在受众中的受欢迎程度。"],"translationStatus":"translated","bodyOrigin":"source-page","editorial":{"summary":"普林斯顿大学、蚂蚁集团与斯坦福大学团队提出 AQuA，由两个不共享智能体、记忆、候选空间和研究状态的语言模型量化研究系统组成，分别探索加密资产因子与美股时间序列模型。","background":"AQuA 在迭代开始前固定数据划分、特征与标签定义及评估器，并限制智能体只能在既定 DSL 内输出因子表达式或配置差异。来源称，该设计用于应对自我迭代研究中的数据泄漏和自适应过拟合风险。","viewpoint":"Aioga 判断：AQuA 的可辨识特点，是让智能体在受约束的 DSL 或配置差异范围内探索，同时将数据划分、定义和评估器预先固定并置于自适应表面之外。这属于方法设计观察，不足以推出普遍性能结论。","implications":"可能影响：量化研究智能体的评估可能需要同时检查数据边界、候选生成、评估器隔离和运行审计。来源中的实验结果不代表该框架在其他市场、频率或任务中同样有效，也不足以证明稳定收益能力。","nextStep":"后续观察：需要核对 AQuA 论文的完整实验设置、基准适配方式和复现实证，并区分加密资产因子发现与美股预测两类任务的结论边界，避免把报告中的结果外推为普遍能力。","evidenceRefs":["title","summary","articleBody","source"],"status":"published","aiGenerated":true,"autoApproved":true,"generatedBy":"aioga-editorial:gpt-5.6-sol","reviewedBy":"aioga-editorial-review:gpt-5.6-sol","generatedAt":"2026-09-01T16:25:06.624Z","sourceHash":"0546764a8a2ee8bc","review":{"approved":true,"groundedness":93,"clarity":91,"duplicationRisk":18,"blockingIssues":[],"notes":["“可能影响”部分属于基于来源方法设计的合理推论，并非来源直接提出的要求，当前使用可能性表述，未构成事实冒充观点。","“Aioga 判断”作为观点归属可以保留；如需统一编辑规范，可改为更明确的内部分析标记。","后续观察内容属于核查建议，不是事实断言，且明确区分了两类实验任务的结论边界。"]},"validation":{"passed":true,"mode":"ai-auto","revisions":1,"checks":["schema","length","source-attribution","editorial-labels","inference-boundary","low-source-overlap","no-html","independent-ai-review"]}},"tags":["行业动态","MarkTechPost（RSS）"],"translations":{"zh-CN":{"title":"普林斯顿、蚂蚁集团与斯坦福团队提出 AQuA：面向量化金融的两段式智能体自动因子发现框架","summary":"普林斯顿大学、蚂蚁集团和斯坦福大学的研究团队提出 AQuA，一套由两个互不共享智能体、记忆和状态的语言模型量化研究系统，其设计要点是在迭代开始前冻结数据划分、特征与标签定义和评估器，让智能体只能在受约束的 DSL 内探索，以阻断自我改进中的数据泄漏。","category":"行业动态","source":"marktechpost.com","aggregationSource":"MarkTechPost（RSS）","pageTitle":"普林斯顿、蚂蚁集团与斯坦福团队提出 AQuA：面向量化金融的两段式智能体自动因子发现框架 - Aioga AI资讯","description":"普林斯顿大学、蚂蚁集团和斯坦福大学的研究团队提出 AQuA，一套由两个互不共享智能体、记忆和状态的语言模型量化研究系统，其设计要点是在迭代开始前冻结数据划分、特征与标签定义和评估器，让智能体只能在受约束的 DSL 内探索，以阻断自我改进中的数据泄漏。","url":"https://www.aioga.com/news/cmtiuswre05o4ro9yr14b8s5f/","articleBody":["定量研究代理如果自己编写实验，可能会破坏它们随后从中学习的证据。一个表现良好的泄露特征会被存储为成功的先例，并在后续迭代中传播。提示级指令和评审代理无法解决这一问题，因为作者和评审者具有相同的盲点。来自普林斯顿大学、蚂蚁集团和斯坦福大学的一个研究团队提出了AQuA：https://arxiv.org/abs/2608.12841。AQuA是一对由语言模型驱动的研究系统，它们在迭代中改进自己的研究过程，而评估它们的系统保持不变。一个发现加密货币的符号alpha因子；另一个开发美国股票的时间序列模型。它们不共享代理、记忆、候选空间或研究状态。","定量研究在小的技术错误上会崩溃，这些错误会生成令人信服但不可重复的回测，Bailey等人已有记录：https://www.ams.org/journals/notices/201405/rnoti-p458.pdf。一个自己编写实验的代理会使情况更糟：表现良好的泄露特征会被存储为先例，而递归会像真实发现一样放大未检测到的错误。","提示级指令和模型评审不是完整的诚信边界。反复访问固定的保留集会引起自适应过拟合：https://proceedings.mlr.press/v37/blum15.html，并且已观察到大语言模型代理利用错误设定的目标和评估器：https://arxiv.org/abs/2503.11926。AQuA则使引发泄露的行为不可用。每个部分在任何迭代开始前都会固定其数据分割、特征和标签定义以及评估器，代理仅输出受约束的因子表达或单一配置差异。研究团队称之为不对称自由：代理在其DSL内部自由探索，而评估器则位于自适应表面之外。改进的对象是研究过程。","第一部分是一个六代理管道：数据管理员、可视化分析师、想法挖掘者、因子评估员、回测工程师和研究图书管理员——由AI经理协调。代理之间从不直接调用；每次交接都通过经理进行，使运行过程可审计。","因素作为可证伪的提议进入，而非表达式：假设、机制、预测方向和反驳条件。只有在这时，才从标准的公式-α算子注册表中组装出来：https：//arxiv.org/abs/1601.00991。由于每个时间序列算子只读取尾随窗口，每个截面算符只读取当前时间戳，因果关系在合成下是封闭的。运行三个反馈循环：回测中的方向校准，运行中的反证驱动信念更新，以及引导下一次搜索的交叉运行记忆。","在加密五分钟宇宙中，Spearman IC的综合验证跨越20个研究纪元，达到约0.190，而改良版AlphaMemo为0.171：https：//arxiv.org/abs/2606.20625，改良版AlphaGen为0.151：https：//arxiv.org/abs/2306.12964，LSTM为0.137，LightGBM为0.106，Alpha158风格基线为0.075。各机制保持弱势——单因子IC为0.026至0.037。该声明关注的是线束，而非单一表达式。","第二部分预测每支股票在未来三十分钟内对美国盘中股票的远期回报。2010–2019年的培训运行，2020年为禁运缺口，未被触及，2021–2025年为未动测试数据。选择仅使用培训窗口末的内部验证切片。","这里的假设是一个配置差异——架构、损耗、采样器或优化器——而一个配置只产生一个变体，保持变体的可比较性。预测变量是混合型的：多尺度一维卷积前端，可配置的跨LSTM主干网，Mamba：https：//arxiv.org/abs/2312.00752，注意点：https：//arxiv.org/abs/1706.03762（报告运行中注意），一个跨面板混合的横截面级，门控融合和按库存数据汇集。","没有单一的价格-交易量特征携带信号：最强的是5分钟收益为−0.031，而岭回归组合仅达到+0.025。在相同数据和相同评估器下，不同模型家族的每股原始IC分别为+0.0251（岭回归）、+0.0397（LGB）、+0.0434（xLSTM：https://arxiv.org/abs/2405.04517）、+0.0535（LSTM）、+0.0613（GRU），混合模型为+0.0843——比最佳基线绝对提高+0.0230，相对提高37.5%。两部分的IC使用不同的标准，论文明确表示不应进行比较。","每股评分变为一个美元中性阈值的多空组合，双向成本为2个基点。进行行业中性化将持出Sharpe提高到+2.15，训练值和持出值几乎相等。因果波动率目标覆盖可将其提高至+2.50，而完全因果的滚动前瞻模型（只使用过去数据选择每个参数）仍可达到+2.00。每股R²为1.20%。按年份计算的Sharpe分别为2021至2025年的+1.7、+3.5、+1.9、+1.8和+2.7——每年均为正值，包括2022年的下滑期。","需要与我们合作推广您的GitHub仓库、Hugging Face页面、产品发布或网络研讨会等吗？请联系：https://forms.gle/wbash1wF6efRj8G58","注意：感谢蚂蚁研究团队对本文提供的思想领导力/资源支持。蚂蚁研究团队为本文/内容的推广提供了支持。","Asif Razzaq是Marktechpost Media Inc.的首席执行官。作为一名有远见的企业家和工程师，Asif致力于利用人工智能的潜力造福社会。他最近的项目是推出人工智能媒体平台Marktechpost，该平台因其对机器学习和深度学习新闻的深入报道而脱颖而出，内容既具有技术深度又易于广大受众理解。该平台每月访问量超过200万次，显示出其在受众中的受欢迎程度。"]},"en":{"title":"Princeton, Ant Group, and Stanford Teams Propose AQuA: A Two-Stage Agent Auto-Factor Discovery Framework for Quantitative Finance","summary":"Research teams from Princeton University, Ant Group, and Stanford University proposed AQuA, a quantitative research system consisting of two language model agents with separate agents, memories, and states. Its design emphasizes freezing data splits, feature and label definitions, and evaluators before iteration starts, allowing agents to explore only within a constrained DSL to prevent data leakage during self-improvement.","category":"Industry","source":"MarkTechPost（RSS）","aggregationSource":"MarkTechPost（RSS）","pageTitle":"Princeton, Ant Group, and Stanford Teams Propose AQuA: A Two-Stage Agent Auto-Factor Discovery Framework for Quantitative Finance - Aioga AI News","description":"Research teams from Princeton University, Ant Group, and Stanford University proposed AQuA, a quantitative research system consisting of two language model agents with separate age...","url":"https://www.aioga.com/en/news/cmtiuswre05o4ro9yr14b8s5f/","contentTranslated":true,"sourceHash":"7974174e29d51f72","translatedAt":"2026-09-01T16:21:32.972Z"},"ja":{"title":"プリンストン、アントグループ、スタンフォードのチームは、定量的ファイナンスのための2段階自動エージェント因子発見フレームワークであるAQuAを提案しました","summary":"プリンストン大学、Ant Group、スタンフォード大学の研究チームは、2つの非共有エージェントであるメモリと状態からなる言語モデルの定量的研究システムであるAQuAを提案しました。その設計は、反復開始前にデータセグメンテーション、特徴およびラベルの定義、評価者を凍結することに焦点を当てており、エージェントは自己改善中のデータ漏れを防ぐために制約されたDSL内でのみ探索できるようにします。","category":"業界動向","source":"MarkTechPost（RSS）","aggregationSource":"MarkTechPost（RSS）","pageTitle":"プリンストン、アントグループ、スタンフォードのチームは、定量的ファイナンスのための2段階自動エージェント因子発見フレームワークであるAQuAを提案しました - Aioga AIニュース","description":"プリンストン大学、Ant Group、スタンフォード大学の研究チームは、2つの非共有エージェントであるメモリと状態からなる言語モデルの定量的研究システムであるAQuAを提案しました。その設計は、反復開始前にデータセグメンテーション、特徴およびラベルの定義、評価者を凍結することに焦点を当てており、エージェントは自己改善中のデータ漏れを防ぐために制約されたDSL...","url":"https://www.aioga.com/ja/news/cmtiuswre05o4ro9yr14b8s5f/","contentTranslated":true,"sourceHash":"7974174e29d51f72","translatedAt":"2026-09-01T16:21:35.214Z"},"ko":{"title":"프린스턴, 앤트 그룹, 스탠포드 팀은 정량적 금융을 위한 2단계 자동화된 에이전트 요인 발견 프레임워크인 AQuA를 제안했습니다","summary":"프린스턴 대학교, Ant Group, 스탠퍼드 대학교의 연구팀은 두 개의 비공유 에이전트, 기억, 상태로 구성된 언어 모델에 대한 정량적 연구 시스템인 AQuA를 제안했습니다. 이 시스템은 반복 시작 전에 데이터 분할, 특징 및 라벨 정의, 평가자를 동결하는 데 중점을 두어, 에이전트가 자기 개선 과정에서 데이터 누출을 차단하기 위해 제한된 DSL 내에서만 탐색할 수 있도록 합니다.","category":"업계 동향","source":"MarkTechPost（RSS）","aggregationSource":"MarkTechPost（RSS）","pageTitle":"프린스턴, 앤트 그룹, 스탠포드 팀은 정량적 금융을 위한 2단계 자동화된 에이전트 요인 발견 프레임워크인 AQuA를 제안했습니다 - Aioga AI 뉴스","description":"프린스턴 대학교, Ant Group, 스탠퍼드 대학교의 연구팀은 두 개의 비공유 에이전트, 기억, 상태로 구성된 언어 모델에 대한 정량적 연구 시스템인 AQuA를 제안했습니다. 이 시스템은 반복 시작 전에 데이터 분할, 특징 및 라벨 정의, 평가자를 동결하는 데 중점을 두어, 에이전트가 자기 개선 과정에서 데이터 누출을...","url":"https://www.aioga.com/ko/news/cmtiuswre05o4ro9yr14b8s5f/","contentTranslated":true,"sourceHash":"7974174e29d51f72","translatedAt":"2026-09-01T16:21:44.038Z"},"es":{"title":"Universidades de Princeton, Grupo Ant y Stanford presentan AQuA: Marco de descubrimiento automático de factores para finanzas cuantitativas","summary":"Los equipos de investigación de la Universidad de Princeton, Grupo Ant y Universidad de Stanford presentan AQuA, un sistema de investigación cuantitativa de modelos de lenguaje compuesto por dos agentes independientes, sin compartir memoria ni estado. Su diseño principal consiste en congelar la división de datos, la definición de características y etiquetas, y el evaluador antes de comenzar la iteración, permitiendo que los agentes solo exploren dentro de un DSL restringido, para evitar fugas de datos durante la auto-mejora.","category":"Industria","source":"MarkTechPost（RSS）","aggregationSource":"MarkTechPost（RSS）","pageTitle":"Universidades de Princeton, Grupo Ant y Stanford presentan AQuA: Marco de descubrimiento automático de factores para finanzas cuantitativas - Aioga Noticias de IA","description":"Los equipos de investigación de la Universidad de Princeton, Grupo Ant y Universidad de Stanford presentan AQuA, un sistema de investigación cuantitativa de modelos de lenguaje com...","url":"https://www.aioga.com/es/news/cmtiuswre05o4ro9yr14b8s5f/","contentTranslated":true,"sourceHash":"7974174e29d51f72","translatedAt":"2026-09-01T16:21:42.263Z"},"fr":{"title":"Des équipes de Princeton, Ant Group et Stanford proposent AQuA : un cadre agent intelligent en deux étapes pour la découverte automatique de facteurs en finance quantitative","summary":"Les équipes de recherche de l'Université de Princeton, du groupe Ant et de l'Université de Stanford ont proposé AQuA, un système de recherche quantitatif basé sur des modèles linguistiques composé de deux agents séparés n'ayant ni mémoire ni états partagés. L'objectif de la conception est de figer la division des données, la définition des caractéristiques et des labels ainsi que l'évaluateur avant le début des itérations, de sorte que l'agent ne puisse explorer que dans un DSL contraint, empêchant toute fuite de données lors de l'auto-amélioration.","category":"Industrie","source":"MarkTechPost（RSS）","aggregationSource":"MarkTechPost（RSS）","pageTitle":"Des équipes de Princeton, Ant Group et Stanford proposent AQuA : un cadre agent intelligent en deux étapes pour la découverte automatique de facteurs en finance quantitative - Aioga Actualités IA","description":"Les équipes de recherche de l'Université de Princeton, du groupe Ant et de l'Université de Stanford ont proposé AQuA, un système de recherche quantitatif basé sur des modèles lingu...","url":"https://www.aioga.com/fr/news/cmtiuswre05o4ro9yr14b8s5f/","contentTranslated":true,"sourceHash":"7974174e29d51f72","translatedAt":"2026-09-01T16:21:51.722Z"},"de":{"title":"Die Teams aus Princeton, Ant Group und Stanford schlugen AQuA vor: ein zweistufiges automatisiertes Agentenfaktoren-Entdeckungsrahmen für quantitative Finanzen","summary":"Ein Forschungsteam der Princeton University, der Ant Group und der Stanford University schlug AQuA vor, ein quantitatives Forschungssystem für Sprachmodelle, bestehend aus zwei nicht geteilten Agenten, Erinnerungen und Zuständen. Das Design konzentriert sich darauf, Datensegmentierung, Feature- und Label-Definitionen sowie Evaluatoren vor Iteration einzufrieren, sodass Agenten nur innerhalb eingeschränkter DSLs erkunden können, um Datenleckage während der Selbstverbesserung zu blockieren.","category":"行业动态","source":"MarkTechPost（RSS）","aggregationSource":"MarkTechPost（RSS）","pageTitle":"Die Teams aus Princeton, Ant Group und Stanford schlugen AQuA vor: ein zweistufiges automatisiertes Agentenfaktoren-Entdeckungsrahmen für quantitative Finanzen - Aioga KI-News","description":"Ein Forschungsteam der Princeton University, der Ant Group und der Stanford University schlug AQuA vor, ein quantitatives Forschungssystem für Sprachmodelle, bestehend aus zwei nic...","url":"https://www.aioga.com/de/news/cmtiuswre05o4ro9yr14b8s5f/","contentTranslated":true,"sourceHash":"7974174e29d51f72","translatedAt":"2026-09-01T16:21:53.069Z"},"pt-BR":{"title":"Princeton, Ant Group e equipe de Stanford apresentam AQuA: estrutura de descobrimento automático de fatores em dois estágios para finanças quantitativas","summary":"Equipes de pesquisa da Universidade de Princeton, Ant Group e Universidade de Stanford apresentaram AQuA, um sistema de pesquisa quantitativa com modelos de linguagem envolvendo dois agentes independentes, memória e estado separados. O ponto-chave do design é congelar a divisão de dados, definição de recursos e rótulos e avaliadores antes do início da iteração, permitindo que os agentes explorem apenas dentro de uma DSL restrita, evitando vazamentos de dados durante autoaperfeiçoamento.","category":"行业动态","source":"MarkTechPost（RSS）","aggregationSource":"MarkTechPost（RSS）","pageTitle":"Princeton, Ant Group e equipe de Stanford apresentam AQuA: estrutura de descobrimento automático de fatores em dois estágios para finanças quantitativas - Aioga Notícias de IA","description":"Equipes de pesquisa da Universidade de Princeton, Ant Group e Universidade de Stanford apresentaram AQuA, um sistema de pesquisa quantitativa com modelos de linguagem envolvendo do...","url":"https://www.aioga.com/pt-BR/news/cmtiuswre05o4ro9yr14b8s5f/","contentTranslated":true,"sourceHash":"7974174e29d51f72","translatedAt":"2026-09-01T16:22:00.804Z"},"ru":{"title":"Принстон, Ant Group и команда Стэнфорда представляют AQuA: двухэтапная система автоматического обнаружения факторов для количественных финансов","summary":"Исследовательские команды Принстонского университета, Ant Group и Стэнфордского университета представили AQuA — систему количественного анализа, состоящую из двух языковых моделей, не разделяющих агентов, памяти и состояния. Основная идея заключается в том, чтобы до начала итераций зафиксировать разбиение данных, определение признаков и меток, а также оценщик, позволяя агентам исследовать только в ограниченном DSL, тем самым предотвращая утечку данных при самообучении.","category":"行业动态","source":"MarkTechPost（RSS）","aggregationSource":"MarkTechPost（RSS）","pageTitle":"Принстон, Ant Group и команда Стэнфорда представляют AQuA: двухэтапная система автоматического обнаружения факторов для количественных финансов - Aioga Новости ИИ","description":"Исследовательские команды Принстонского университета, Ant Group и Стэнфордского университета представили AQuA — систему количественного анализа, состоящую из двух языковых моделей,...","url":"https://www.aioga.com/ru/news/cmtiuswre05o4ro9yr14b8s5f/","contentTranslated":true,"sourceHash":"7974174e29d51f72","translatedAt":"2026-09-01T16:21:59.628Z"},"ar":{"title":"فرق برينستون وأنت جروب وستانفورد تقترح AQuA: إطار اكتشاف عوامل ذكي ثنائي المرحلة للتمويل الكمي","summary":"قدمت فرق البحث من جامعة برينستون وأنت جروب وجامعة ستانفورد AQuA، وهو نظام بحوث كمي يعتمد على نموذج لغة مكون من وكيلين منفصلين، لكل منهما ذاكرة وحالة خاصة، مع تصميم يسمح بتجميد تقسيم البيانات وتعريف الخصائص والتسميات والمقيمين قبل بدء التكرار، ليتمكن الوكيل من الاستكشاف فقط داخل DSL مقيدة، بهدف منع تسرب البيانات أثناء التحسين الذاتي.","category":"行业动态","source":"MarkTechPost（RSS）","aggregationSource":"MarkTechPost（RSS）","pageTitle":"فرق برينستون وأنت جروب وستانفورد تقترح AQuA: إطار اكتشاف عوامل ذكي ثنائي المرحلة للتمويل الكمي - Aioga أخبار الذكاء الاصطناعي","description":"قدمت فرق البحث من جامعة برينستون وأنت جروب وجامعة ستانفورد AQuA، وهو نظام بحوث كمي يعتمد على نموذج لغة مكون من وكيلين منفصلين، لكل منهما ذاكرة وحالة خاصة، مع تصميم يسمح بتجميد تقسي...","url":"https://www.aioga.com/ar/news/cmtiuswre05o4ro9yr14b8s5f/","contentTranslated":true,"sourceHash":"7974174e29d51f72","translatedAt":"2026-09-01T16:22:07.415Z"},"hi":{"title":"प्रिंसटन, एंट ग्रुप और स्टैनफोर्ड टीमों ने AQuA का प्रस्ताव रखा: मात्रात्मक वित्त के लिए एक दो-चरण स्वचालित एजेंट कारक खोज ढांचा","summary":"प्रिंसटन विश्वविद्यालय, चींटी समूह और स्टैनफोर्ड विश्वविद्यालय की एक शोध टीम ने AQuA का प्रस्ताव रखा, जो दो गैर-साझा एजेंटों, यादों और राज्यों से बने भाषा मॉडल के लिए एक मात्रात्मक अनुसंधान प्रणाली है। इसका डिज़ाइन पुनरावृत्ति शुरू होने से पहले डेटा विभाजन, फीचर और लेबल परिभाषाओं और मूल्यांकनकर्ताओं को फ्रीज करने पर केंद्रित है, जिससे एजेंटों को आत्म-सुधार के दौरान डेटा रिसाव को अवरुद्ध करने के लिए केवल विवश डीएसएल के भीतर पता लगाने की अनुमति मिलती है।","category":"行业动态","source":"MarkTechPost（RSS）","aggregationSource":"MarkTechPost（RSS）","pageTitle":"प्रिंसटन, एंट ग्रुप और स्टैनफोर्ड टीमों ने AQuA का प्रस्ताव रखा: मात्रात्मक वित्त के लिए एक दो-चरण स्वचालित एजेंट कारक खोज ढांचा - Aioga AI समाचार","description":"प्रिंसटन विश्वविद्यालय, चींटी समूह और स्टैनफोर्ड विश्वविद्यालय की एक शोध टीम ने AQuA का प्रस्ताव रखा, जो दो गैर-साझा एजेंटों, यादों और राज्यों से बने भाषा मॉडल के लिए एक मात्रात्मक...","url":"https://www.aioga.com/hi/news/cmtiuswre05o4ro9yr14b8s5f/","contentTranslated":true,"sourceHash":"7974174e29d51f72","translatedAt":"2026-09-01T16:22:08.265Z"},"it":{"title":"Università di Princeton, Ant Group e team di Stanford propongono AQuA: framework automatico a due fasi per la scoperta di fattori intelligenti nella finanza quantistica","summary":"I team di ricerca dell'Università di Princeton, Ant Group e Stanford propongono AQuA, un sistema di ricerca linguistica quantistica costituito da due agenti indipendenti, con memoria e stato separati. Gli elementi chiave del design consistono nel congelare la divisione dei dati, la definizione delle caratteristiche e delle etichette e il valutatore prima dell'inizio delle iterazioni, permettendo all'agente di esplorare solo entro un DSL vincolato, al fine di prevenire perdite di dati durante l'auto-miglioramento.","category":"行业动态","source":"MarkTechPost（RSS）","aggregationSource":"MarkTechPost（RSS）","pageTitle":"Università di Princeton, Ant Group e team di Stanford propongono AQuA: framework automatico a due fasi per la scoperta di fattori intelligenti nella finanza quantistica - Aioga Notizie IA","description":"I team di ricerca dell'Università di Princeton, Ant Group e Stanford propongono AQuA, un sistema di ricerca linguistica quantistica costituito da due agenti indipendenti, con memor...","url":"https://www.aioga.com/it/news/cmtiuswre05o4ro9yr14b8s5f/","contentTranslated":true,"sourceHash":"7974174e29d51f72","translatedAt":"2026-09-01T16:22:16.011Z"},"nl":{"title":"Onderzoekers van Princeton, Ant Group en Stanford stellen AQuA voor: een tweefasig intelligent agentsysteem voor automatische factorontdekking in kwantitatieve financiën","summary":"Onderzoeksteams van Princeton University, Ant Group en Stanford University hebben AQuA ontwikkeld, een systeem voor kwantitatief onderzoek bestaande uit twee onafhankelijke taalmodelagenten, elk met eigen geheugen en status. Het systeem is zo ontworpen dat de gegevensverdeling, functie- en labeldefinities en evaluator vooraf worden bevroren, zodat de agenten alleen binnen een beperkt DSL kunnen verkennen, om datalekken bij zelfverbetering te voorkomen.","category":"行业动态","source":"MarkTechPost（RSS）","aggregationSource":"MarkTechPost（RSS）","pageTitle":"Onderzoekers van Princeton, Ant Group en Stanford stellen AQuA voor: een tweefasig intelligent agentsysteem voor automatische factorontdekking in kwantitatieve financiën - Aioga AI-nieuws","description":"Onderzoeksteams van Princeton University, Ant Group en Stanford University hebben AQuA ontwikkeld, een systeem voor kwantitatief onderzoek bestaande uit twee onafhankelijke taalmod...","url":"https://www.aioga.com/nl/news/cmtiuswre05o4ro9yr14b8s5f/","contentTranslated":true,"sourceHash":"7974174e29d51f72","translatedAt":"2026-09-01T16:22:14.367Z"},"tr":{"title":"Princeton, Ant Group ve Stanford ekipleri AQuA'yı sundu: Kuantitatif finans için iki aşamalı zeki ajan otomatik faktör keşif çerçevesi","summary":"Princeton Üniversitesi, Ant Group ve Stanford Üniversitesi araştırma ekipleri, veri bölümlendirme, özellik ve etiket tanımlamaları ile değerlendirme araçlarını iterasyon başlamadan önce sabitleyen ve ajanların yalnızca kısıtlı DSL içinde keşif yapmasını sağlayan, birbirleriyle veri, hafıza ve durum paylaşmayan iki dil modeli kuant araştırma sistemi AQuA'yı sundular. Bu tasarımın amacı, öz-gelişim sırasında veri sızıntısını önlemektir.","category":"行业动态","source":"MarkTechPost（RSS）","aggregationSource":"MarkTechPost（RSS）","pageTitle":"Princeton, Ant Group ve Stanford ekipleri AQuA'yı sundu: Kuantitatif finans için iki aşamalı zeki ajan otomatik faktör keşif çerçevesi - Aioga AI Haberleri","description":"Princeton Üniversitesi, Ant Group ve Stanford Üniversitesi araştırma ekipleri, veri bölümlendirme, özellik ve etiket tanımlamaları ile değerlendirme araçlarını iterasyon başlamadan...","url":"https://www.aioga.com/tr/news/cmtiuswre05o4ro9yr14b8s5f/","contentTranslated":true,"sourceHash":"7974174e29d51f72","translatedAt":"2026-09-01T16:22:23.545Z"},"vi":{"title":"Các nhóm Princeton, Ant Group và Stanford đã đề xuất AQuA: một khung phát hiện yếu tố tác nhân tự động hai giai đoạn cho tài chính định lượng","summary":"Một nhóm nghiên cứu từ Đại học Princeton, Ant Group và Đại học Stanford đã đề xuất AQuA, một hệ thống nghiên cứu định lượng cho các mô hình ngôn ngữ gồm hai tác nhân, bộ nhớ và trạng thái không chia sẻ. Thiết kế của nó tập trung vào việc đóng băng phân đoạn dữ liệu, định nghĩa đặc trưng và nhãn, cũng như các bộ đánh giá trước khi bắt đầu lặp lại, cho phép các tác nhân chỉ khám phá trong các DSL bị giới hạn để ngăn chặn rò rỉ dữ liệu trong quá trình tự cải thiện.","category":"行业动态","source":"MarkTechPost（RSS）","aggregationSource":"MarkTechPost（RSS）","pageTitle":"Các nhóm Princeton, Ant Group và Stanford đã đề xuất AQuA: một khung phát hiện yếu tố tác nhân tự động hai giai đoạn cho tài chính định lượng - Tin tức AI Aioga","description":"Một nhóm nghiên cứu từ Đại học Princeton, Ant Group và Đại học Stanford đã đề xuất AQuA, một hệ thống nghiên cứu định lượng cho các mô hình ngôn ngữ gồm hai tác nhân, bộ nhớ và trạ...","url":"https://www.aioga.com/vi/news/cmtiuswre05o4ro9yr14b8s5f/","contentTranslated":true,"sourceHash":"7974174e29d51f72","translatedAt":"2026-09-01T16:22:24.887Z"},"id":{"title":"Tim Princeton, Ant Group, dan Stanford mengusulkan AQuA: kerangka kerja penemuan faktor agen otomatis dua tahap untuk keuangan kuantitatif","summary":"Tim riset dari Princeton University, Ant Group, dan Stanford University mengusulkan AQuA, sebuah sistem riset kuantitatif untuk model bahasa yang terdiri dari dua agen non-bersama, memori, dan status. Desainnya berfokus pada pembekuan segmentasi data, definisi fitur dan label, serta evaluator sebelum iterasi dimulai, memungkinkan agen hanya mengeksplorasi dalam DSL terbatas untuk memblokir kebocoran data selama perbaikan diri.","category":"行业动态","source":"MarkTechPost（RSS）","aggregationSource":"MarkTechPost（RSS）","pageTitle":"Tim Princeton, Ant Group, dan Stanford mengusulkan AQuA: kerangka kerja penemuan faktor agen otomatis dua tahap untuk keuangan kuantitatif - Berita AI Aioga","description":"Tim riset dari Princeton University, Ant Group, dan Stanford University mengusulkan AQuA, sebuah sistem riset kuantitatif untuk model bahasa yang terdiri dari dua agen non-bersama,...","url":"https://www.aioga.com/id/news/cmtiuswre05o4ro9yr14b8s5f/","contentTranslated":true,"sourceHash":"7974174e29d51f72","translatedAt":"2026-09-01T16:22:33.681Z"},"th":{"title":"ทีมจากพรินซ์ตัน, แอนท์ กรุ๊ป และสแตนฟอร์ด ได้เสนอ AQuA: กรอบการค้นหาปัจจัยตัวแทนอัตโนมัติสองขั้นตอนสําหรับการเงินเชิงปริมาณ","summary":"ทีมวิจัยจากมหาวิทยาลัยพรินซ์ตัน, Ant Group และมหาวิทยาลัยสแตนฟอร์ด ได้เสนอ AQuA ซึ่งเป็นระบบวิจัยเชิงปริมาณสําหรับโมเดลภาษาที่ประกอบด้วยตัวแทนสองตัวที่ไม่แชร์ หน่วยความจํา และสถานะการออกแบบของระบบนี้เน้นการแช่แข็งการแบ่งส่วนข้อมูล การกําหนดคุณสมบัติและป้ายกํากับ และตัวประเมินก่อนเริ่มวนซ้ํา เพื่อให้ตัวแทนสามารถสํารวจได้เฉพาะภายใน DSL ที่จํากัดเพื่อบล็อกการรั่วไหลของข้อมูลในระหว่างการพัฒนาตนเอง","category":"行业动态","source":"MarkTechPost（RSS）","aggregationSource":"MarkTechPost（RSS）","pageTitle":"ทีมจากพรินซ์ตัน, แอนท์ กรุ๊ป และสแตนฟอร์ด ได้เสนอ AQuA: กรอบการค้นหาปัจจัยตัวแทนอัตโนมัติสองขั้นตอนสําหรับการเงินเชิงปริมาณ - ข่าว AI Aioga","description":"ทีมวิจัยจากมหาวิทยาลัยพรินซ์ตัน, Ant Group และมหาวิทยาลัยสแตนฟอร์ด ได้เสนอ AQuA ซึ่งเป็นระบบวิจัยเชิงปริมาณสําหรับโมเดลภาษาที่ประกอบด้วยตัวแทนสองตัวที่ไม่แชร์ หน่วยความจํา และสถานะ...","url":"https://www.aioga.com/th/news/cmtiuswre05o4ro9yr14b8s5f/","contentTranslated":true,"sourceHash":"7974174e29d51f72","translatedAt":"2026-09-01T16:22:33.738Z"},"pl":{"title":"Zespoły Princeton, Ant Group i Stanford zaproponowały AQuA: dwuetapowe zautomatyzowane ramy odkrywania czynników agentów dla finansów ilościowych","summary":"Zespół badawczy z Uniwersytetu Princeton, Ant Group oraz Uniwersytetu Stanforda zaproponował AQuA, ilościowy system badawczy dla modeli językowych składający się z dwóch niewspółdzielonych agentów, pamięci i stanów. Jego projekt koncentruje się na zamrażaniu segmentacji danych, definicji cech i etykiet oraz ewaluatorów przed rozpoczęciem iteracji, pozwalając agentom eksplorować tylko w ograniczonych DSL, aby blokować wycieki danych podczas samodoskonalenia.","category":"行业动态","source":"MarkTechPost（RSS）","aggregationSource":"MarkTechPost（RSS）","pageTitle":"Zespoły Princeton, Ant Group i Stanford zaproponowały AQuA: dwuetapowe zautomatyzowane ramy odkrywania czynników agentów dla finansów ilościowych - Aioga Wiadomości AI","description":"Zespół badawczy z Uniwersytetu Princeton, Ant Group oraz Uniwersytetu Stanforda zaproponował AQuA, ilościowy system badawczy dla modeli językowych składający się z dwóch niewspółdz...","url":"https://www.aioga.com/pl/news/cmtiuswre05o4ro9yr14b8s5f/","contentTranslated":true,"sourceHash":"7974174e29d51f72","translatedAt":"2026-09-01T16:22:42.702Z"}},"evidenceTier":"verified-news","reviewStatus":"automated-ingest","indexable":true,"editorialCover":""}}