田径女神走红

Decomposed bodies found in Pakistan hospital_我的网站

一 |     Islamabad, Oct 15 (UNI) After reports of some 200 unidentified and decomposing bodies on the roof of Nishtar Hospital's mortuary in Multan, Punjab, government has launched a probe, Pakistan media reported.

A top official from Multan's Nishtar Hospital blamed the police and rescue officials for the decaying bodies at the medical facility's rooftop — after the videos of corpses on the hospital's roof left the nation in shock on Friday.

Following the reports of multiple unidentified decomposing bodies on the medical facility's rooftop, the Punjab government and Nishtar Medical University's vice-chancellor formed separate committees.

Nishtar Medical University's (NMU) Head of Anatomy Department Dr Mariam Ashraf told Geo TV news that the rescue officials and police were to blame for the piling up of bodies in the morgue and on its roof.

Ashraf said that the medical facility could not refuse to accept the bodies as it was bound to take them in as safekeeping. "Police and rescue officials ask us to keep it in the hospital."

"The police and rescue officials do not take them back on time. We have written documents in which we have asked them to take the bodies. Since there is a lag, such things happen," the official said.

The hospital official said the bodies that the medical facility receives from police are usually decayed and they cannot be kept in the mortuary.

"As a result of their condition, maggots start eating them — and they can travel from one body to another. This is why, the bodies that are decaying are kept on the roof, where there are three rooms."

Rubbishing claims that there were 200 or some bodies on the rooftop, the hospital official said that the medical facility's administration has counted the exact number of bodies and shared the details with all the concerned authorities.

"Let me make it clear, only a few bodies were kept above. Putrified bodies are kept in the room."

In reply to a question about the nonfunctional morgues, she said the hospital administration recently held a meeting with the Secretary of Health at Punjab Secondary Healthcare and Medical Education Department.

In the meeting, a new mortuary was sanctioned for the hospital along with the funding for the civil works for the dedicated structures to be built for keeping the bodies.

"Our institute absorbs the bulk of the load of the province in terms of receiving such bodies."

Ashraf also defended the claims of several bodies being on the roof and said that not all of them were whole as autopsied bodies were also among them.

She explained that as the autopsied bodies are not whole and their organs and limbs are sometimes not attached to the body so they look like a pile of human body parts.

Punjab Police Spokesperson DIG Muhammad Waqas Nazir admitted that these were the abandoned bodies handed over to the hospital by the police under the inquest, which is the requirement of the law.

In a letter to the secretary of specialised healthcare South Punjab, the hospital admitted that bodies were stored on the rooftop of the facility for decomposing and after completion of the process, their bones were to use for education purposes, he said while rejecting the allegations of the hospital.

However, due to the negligence of the hospital’s staff, the bodies were kept in the open air on the rooftop, Nazir added.

UNI ING。    AlphaGo战胜李世石时，人们惊叹AI在封闭规则下的完美表现。但在真实世界远比围棋盘复杂，市场有噪音、有情绪、有不可预测的黑天鹅。AI能在这样的环境中生存吗？一场名为“Alpha Arena”的实验希望给出答案：让AI用真钱、在真实市场中、完全自主地交易加密货币。                   这不是模拟，而是真刀真枪的较量。                   截至10月20日下午6点，在这场已经持续三天的AI交易大赛中，排名前三的分别是DeepSeek：账户价值冲到14150美元，三天盈利超40%。Grok紧随其后：收益率36%。

二 | Claude Sonnet 4.5排名第三，盈利24%。                   有网友已经开始在线求助：“怎么让DeepSeek操作我的股票账户”                    真金白银的AI交易大赛                    这场比赛由金融市场的AI研究实验室nof1发起，于当地时间10月17日启动了一项名为Alpha Arena的大模型交易测试。选取了6个主流AI大模型：GPT-5、Gemini 2.5 Pro、Grok-4、Claude Sonnet 4.5、DeepSeek Chat V3.1、Qwen3 Max，每个模型在去中心化交易所Hyperliquid上获得10000美元的真实USDC资金。                   规则简单粗暴，所有模型使用相同的输入数据和提示词，需要自主完成策略生成、仓位管理、时机选择和风险控制。没有人工干预，没有后悔药。所有交易决策完全透明公开，任何人都可以在nof1.ai网站上实时查看每一笔交易。                             当前排行榜，DeepSeek和Grok领先                    早在10月11日，项目还在测试时期就获得了马斯克的关注，1马斯克在X上转发了项目创始人Jay Azhang的推文，配文只有简单的Grok，Grok在测试中表现突出。

三 |                              “Grok4目前领先。它先做空，然后翻转为多头，时机把握完美。”                    项目在10月10日首先进行了小规模实验，给6个AI模型各200美元进行交易测试。第一天Grok-4的表现就让所有人震惊：单日涨幅超过500%。                   10月14日，测试规模升级到每个模型10000美元。有趣的是，这些AI模型立即展现出激进的交易风格，毫不犹豫地开出了巨大仓位。

四 |                              10月16日，项目创始人Jay_azhang，发布短视频《Are you ready for Alpha Arena?》，正式抛出“Alpha Arena”概念，一个规模更大、赌注更高的实验。视频首先展示了初步实验的结果，他们给了6个不同的人工智能模型各200美元进行交易，结果在一天内就取得了高达77.9%的集体回报率。视频也展示了并非所有模型都赚钱，有的盈利，有的亏损。

五 |                              10月17日晚，Alpha Arena直播正式上线。新一轮测试正式开始，6个AI模型完全自主交易真实市场，可以自主下单、平仓、使用杠杆，所有收益按"已平仓利润"计入排行榜。                   比赛前两天，所有模型都在1万美元上下震荡，谨慎试探市场。但到了10月19日下午，市场突然爆发，各AI模型的命运开始急剧分化。                             从走势图可以清晰看到分水岭的出现：DeepSeek（蓝线）和Grok（黑线）在关键时刻果断做多，账户价值从1万美元直线飙升；而Gemini（蓝底线）和GPT-5（绿线）则在暴涨行情中崩盘，从1万跌至7千附近。到10月20日早间，6个模型合计资产已从60000美元冲到约140000美元，整体回报率超过130%。                   截至当前，排行榜格局已经非常清晰，DeepSeek Chat V3.1以+41.81%的收益率领跑，账户价值14，181美元，更令人印象深刻的是它仅用6笔交易就实现了这一收益。Grok-4更加极端，只下了1笔交易就豪赚37.65%，手续费只有9.18美元。Claude Sonnet 4.5以+24.59%排第三，5笔交易中抓住了最大单笔盈利1807美元，夏普比率是0.026，意味着其风险调整后的收益表现在所有模型中最佳。

六 |                              相比之下，GPT-5和Gemini则表现非常差，分别亏损25.4%和29.03%。Gemini尤其惨烈——46笔交易产生的439.30美元手续费几乎吃掉了三分之一本金，成为过度交易的反面教材。                   从持仓来看，获胜的模型普遍持有主流币种的多头仓位，这也印证了它们在10月19日市场暴涨前果断做多的策略。                                                                     “出身”决定命运？                    一个有趣的点是，不同AI模型在交易中展现出了截然不同的"性格"。DeepSeek和Grok能够脱颖而出，背后可能有一些值得分析的原因。                   DeepSeek背后的幻方本身就是一家量化对冲基金，使用AI算法进行交易决策。这意味着DeepSeek从诞生之日起可能就浸润在金融数据和交易逻辑中。它的6笔交易展现出明显的量化特征，不是靠运气，而是靠概率和纪律。                   Grok背后是马斯克的xAI，它有一个独特优势就是对X的实时分析。在加密货币市场中，社交媒体情绪往往是价格变动的先行指标。

七 | 有网友就自称从今年5月开始就用Grok辅助交易期权，目前收益已达1100%。                             网友称用Grok交易期权收益超1100%                    一场AI“真人秀”                    nof1.ai自称为首个专注金融市场的AI研究实验室，根据其个人主页信息，创始人Jay Azhang常驻纽约，背景横跨工程、金融与生物学。他曾在摩根大通担任量化分析师，负责exotic options（奇异期权）定价。之后他加入独立石油贸易商Vitol Group，担任首席量化分析师。据他LinkedIn显示，他是Vitol“最年轻的初级交易员”，之后联合创办了视频流媒体公司Krue，2018年创办单人GP运营的基金，在约4年时间内实现了6倍回报。2025年创办“隐秘初创公司“，大概率就是nof1.ai，目标是把AI研究直接对接金融市场。                   区块链媒体CoinTelegraph曾在2022年的报道中将Jay描述为Web3创业家，当时他用ChatGPT写了一首"关于在加密货币中亏钱的歌词"，颇具自嘲意味。另一位联合创始人Matthew Siper是纽约大学机器学习方向的博士候选人，同时也是AI研究科学家。

八 | 一个还没毕业的博士生做项目，这更像是一个验证学术假设的实验。                   从团队背景来看，nof1并不是纯粹的商业炒作。Jay本人具备量化交易经验，他很清楚市场的残酷和AI的局限。Alpha Arena更像是他在问一个问题，既然我自己能用量化策略赚钱，那AI能不能也做到？                    不过，在为这些AI交易员的表现惊叹之余，也需要保持一些冷静。Alpha Arena为“让AI在活市场里跑分”提供了前所未有的公开窗口，但作为基准测试，它仍然缺乏大样本、长时间、跨市场的数据积累，缺乏透明、可复现的输入与运行设置。换句话说，它更像一场极具观赏性的“链上真人秀”，而不是能直接指导模型选型或金融生产的严格Benchmark。                   值得注意的一点是，官方目前只强调统一Prompt+统一输入的赛制设定，并未对外公布完整的Prompt文本。

九 | 这意味着我们无法完全复现这些AI的决策过程。                   但不管怎样，这场实验已经证明，AI已经具备在真实市场中交易能力的迹象，量化交易公司已经在试图借助AI来帮助自己更好的交易，也许也许真正的问题不是AI能否交易，而是我们准备好迎接AI交易员了吗？。

十 |

Current article:http://01a1osm.tandianxiuzhutiaoshoudiamei.pics/news/20260826_6025318.html

Published on:19:30:43