KINFX🇭🇰 (@kinfx_) header
KINFX🇭🇰 (@kinfx_) profile picture

Prop Firm Simulator - Steam Prop firm 連結 → https://nitter.cf/t.co/yQFTmsTOpc Free Journal → https://nitter.cf/t.co/GvcCWG9wjf

Joined March 2025
Photo by KINFX🇭🇰 (@kinfx_)Photo by KINFX🇭🇰 (@kinfx_)Photo by KINFX🇭🇰 (@kinfx_)Photo by KINFX🇭🇰 (@kinfx_)Photo by KINFX🇭🇰 (@kinfx_)Photo by KINFX🇭🇰 (@kinfx_)Photo by KINFX🇭🇰 (@kinfx_)Photo by KINFX🇭🇰 (@kinfx_)Photo by KINFX🇭🇰 (@kinfx_)Photo by KINFX🇭🇰 (@kinfx_)
1/ TPT(Take Profit Trader) PRO 帳戶當日可出金,80/20 分帳;PRO+ 是 90/10 。 碼 KINFX:takeprofittrader.com/?referr… 2/ Tradeify Select Daily 適合要短周期出金的人;彈性方案較貴。 碼 KINFX:tradeify.co/?ref=kinfx 3/ Lucid Trading 考試規則可客製(Daily/Flex 兩種出金節奏),香港可用。 碼 KINFX:lucidtrading.com/ref/2078 4/ Apex Trader Funding 多帳戶、入場門檻低,Legacy 30% MAE 條款要先看懂。 碼 KINFX:apextraderfunding.com/member… Link: linktr.ee/kinfx 華人可玩地區一覽 propfirmsdrama.com/2026/10/2… #自營交易 #PropFitm #期貨
1
2
1,105
KINFX🇭🇰 retweeted
Rockstar Games 周三解雇了我。第二天 GTA 6 就发售了 我是 Rockstar North 的高级引擎工程师,从事 RAGE 性能优化。三个月前,我在遥测模块里发现了一条本不该存在的数据管道,而且我能证明它流向了哪里 此刻每一份正在运行的 GTA 6 都在收集玩家的行为数据。这部分不算新闻。每款游戏都会收集遥测数据。隐私政策里写的是"用于改进未来产品的匿名游戏数据分析"。我发现的不是这个 遥测模块的数据会汇入 Take-Two 内部的一个系统,名为 Lens。Lens 并不会把匿名使用统计数据发送到 Rockstar 开发人员的数据看板。它为每一位玩家建立一份个人行为画像。不是匿名的。与你的主机账号、你的平台 ID、你的 IP 绑定 我在8月14日向我的技术主管报告了这件事。他说这是"用于玩家体验研究的扩展分析",让我别管这事。我要求提供数据处理协议。他说这是在 Take-Two 集团层面处理的,不在 Rockstar 的职责范围之内 我调取了 Lens 数据管道的完整文档。我发现的内容如下: 遥测的内容远远超出游戏时长和崩溃报告。Lens 以毫秒级精度记录每一次输入。按键时间点、摇杆动作、反应速度、PS5 自适应扳机上的扳机压力模式。它实时运行行为分析,并生成他们称之为"玩家行为签名"的东西 - 一份对每位玩家都独一无二的画像,就像指纹对每个人都独一无二一样 Lens 会从你的游戏过程中推导出:冲动控制评分、风险承受度指数、攻击性反应模式、耐心阈值、压力下的决策速度、消费冲动评级。这些不是调试指标。这些是心理行为类别 直到我找到了对外传输的管道,我才明白它们是用来做什么的 这些数据不会留在 Rockstar。也不会留在 Take-Two。Lens 通过内部 API 将行为画像传送给 Take-Two 旗下一家名为 Vantage Analytics 的子公司。Vantage 没有公开网站,没有新闻页面,没有招聘信息。它于2023年在特拉华州注册成立 Vantage 是一家行为 AI 公司。它用玩家画像训练预测模型,并把这些模型授权给商业客户。管道文档中包含一份不完整的客户清单。两家保险集团,一家信用风险建模公司,一个人力资源筛选平台,以及一家广告数据经纪商 你的 GTA 6 游戏过程此刻正在生成一份行为评分。没人追你的时候,你开得有多快。你会不会在红灯前停下。NPC 无端攻击你时,你怎么反应。你行动之前要想多久。你花游戏内的钱有多冲动。每一个选择都在喂养一个模型,企业正在付费查询这个模型,用来对真实的人做决定
3,261
26,630
3,275
150,940
7,071,042
KINFX🇭🇰 retweeted
Nikola Jokic when asked if he will play GTA VI: “No… I play League of Legends on the phone” 📱😅 (via @Shapalicious)
233
1,201
510
32,930
1,890,651
KINFX🇭🇰 retweeted
🚨Warning: Alpha Funded (formerly Alpha Trader Firm) and its ties to the CFTC's Traders Domain case I've been seeing a lot more ads and posts for Alpha Funded lately, so given the history here I feel like this warning needs to go out. To be clear, this is not Alpha Futures or Alpha Capital Group. Those are separate unrelated companies and are not part of this case. Alpha's founder and former majority owner, Alex Santi, remains a defendant in the CFTC's ongoing case, which is currently stayed, and the court-appointed receiver says Alpha paid at least $3.4M toward an AmEx tied to him while his assets were frozen. The original Alpha Trader company is now in a federal receivership, similar to how a bankruptcy trustee took over Fast Track Trading's estate. The difference is the receiver says Alpha kept operating and bringing in millions the whole time, and Alpha Funded is still out there selling challenges today. Who owns Alpha Funded now has never been publicly disclosed. CFTC v. Traders Domain FX Ltd., S.D. Florida, 1:24-cv-23745. 🧵
20
15
6
91
39,434
在 M4 Max 64GB 上實測 Splash 1.7.0(Qwen3.8-27B 4bit),同一模型級別對比 llama.cpp 與 MLX: · 生成 36–46 tok/s(llama.cpp 23/MLX 29) · Prefill 352 tok/s(242/233) · 34.8K prompt 重跑 1.9 秒(prefix cache) · 4 路併發合計 95 tok/s · 256K 上下文、視覺、工具調用、OpenAI 與 Anthropic 雙 API brew install,約 2.5 分鐘就緒。快 1.4–2 倍,感覺不錯,我會等M5 Ultra 送到,再試試其他Model 。 github.com/incoai/splash
1
3
421
想測試自己是否具備足夠知識, 可以去玩玩考試選擇題😂 propfirmsdrama.com/readiness…
1
1
3
1,021
KINFX🇭🇰 retweeted
Got tired of my mouse dying mid-game, so I put the battery % on my taskbar. Works for headsets, controllers & more. Very lightweight, made in C++ and open source. GitHub link and exe in comments
1,077
6,152
1,049
78,578
5,082,871
等T1 官宣 @grok 你覺得誰最大機會嬴?
講中Zeus 續約幾年, 我送一個TPT 50K 帳號!😎 先到先得
1
1
476
KINFX🇭🇰 retweeted
Introducing Firecrawl Local. We rebuilt the Firecrawl repo so agents can scrape locally. - 37x faster rust based scrape primitive (vs self-hosted) - 230x less memory (11 MB idle) - 100% open source Compatible out of the box with all of our MCP & SDKs. github.com/firecrawl/firecra…
239
382
128
5,560
848,332
KINFX🇭🇰 retweeted
Gemini 4 Neon is rolling out now. Sitting between Carbon and Hydrogen, it brings frontier intelligence and fast task completion for complex software engineering, knowledge work, and science.
400
983
680
10,148
1,011,903
KINFX🇭🇰 retweeted
Some of you have been asking very politely. Some of you have just been asking. GPT-6.2 Astra and Sol are now available. Super efficient and built to perform where it counts—the real world. openai.com/index/introducing…
2,292
3,610
3,729
36,029
13,013,008
講中Zeus 續約幾年, 我送一個TPT 50K 帳號!😎 先到先得
34
1
8
1,637
is so over 玩完了
Replying to @JohnLeFevre
Hong Kong has had one of the lowest birth rates in the world for a long time. They will be gone soon if that doesn’t change.
4
792
這個一定要學 核心Skill 被背景改寫之後飄移及瘋狂膨脹 一定要關掉😂
Replying to @JankDankins_
Yes. One config line turns off the automatic self-improvement review and leaves memory working: hermes config set auxiliary.background_review.enabled false (or in ~/.hermes/config.yaml) auxiliary: background_review: enabled: false What stays the same: memory still loads into every session, and the agent can still call the memory tool during a normal turn when it decides to save something. What stops: the background job that runs after turns to review the conversation and write memories or skills. That job is a second model call. On a Spark running a local model it competes with your next message for the same inference server, which is probably why things grind to a halt. It's a config setting, so it applies to every turn and every session. You don't need to set it per turn. If you ever want a review, /refine runs one on demand.
3
469
無聊叫AI 寫, 看看效果 一種靠 Prop Firm 不對稱成本,另一種是自己真倉承擔真正回撤,卻有人自稱為「成功交易員」,承認真倉打不贏並不可恥,最可恥是誤導 買 10 個 50K 帳號花 1,000 美元,爆掉 9 個消耗 18,000 美元 Drawdown,只要第 10 個出金超過 1,000 美元就算賺——這賺到的是「不對稱成本」,不是真倉可行的交易策略。... propfirmsdrama.com/2026/11/1…
1
5
1,022
大家會發現,愈來愈多人在社交媒體上分享自己製作的交易工具,或各種與交易相關的內容。 但我上星期製作的版本,甚至兩個月前的 backtest 工具,其實已經完成了詳細教學。 大家不要相信那些人說「需要寫幾萬行程式碼、調校好幾天、把 Opus 用到極限」這類說法。實際上,兩小時已經可以完成大部分工作;最多也只是花半天慢慢調整細節。😂😂 我分享的方法很簡單、輕鬆,而且有免費方案:選對 harness,再加上幾條 prompt,就能解決問題。整條 YouTube 影片也是這樣製作,沒有花巧的剪接或技術。 有些人分享的方法卻說得很複雜,強調需要花很多時間和心力,不容易做到;然後建議你先報名上課,加入後才會「毫無保留」地教你。 今後我還會分享很多與 AI 相關的無聊內容,也會陸續製作一些實驗性質的專案,部分會相當有趣。
10
912
KINFX🇭🇰 retweeted
Replying to @wholemars
Haven’t even had sex in ages (sigh)
4,788
7,166
7,950
45,370