Garfield | 𝐦𝐞𝐦𝐨𝐫𝐲_𝐨𝐫𝐝𝐞𝐫_𝐚𝐜𝐪_𝐫𝐞𝐥

太空 技術和科學 科學 科技新聞 美國內閣 美國與全球

[Garfield | 𝐦𝐞𝐦𝐨𝐫𝐲_𝐨𝐫𝐝𝐞𝐫_𝐚𝐜𝐪_𝐫𝐞𝐥 @Garfield9888]

@chenchengpro 看到一个博主说，cc和codex架构是两条路线，前者一个session一个bun，后者app-server架构，全端统一runtime，qodercli未来会怎么选啊

[Garfield | 𝐦𝐞𝐦𝐨𝐫𝐲_𝐨𝐫𝐝𝐞𝐫_𝐚𝐜𝐪_𝐫𝐞𝐥 @Garfield9888]

最近研究效率直接翻倍：把几十个资料标签页丢给 @tabbitbrowser，几分钟就能拿到一份结构清晰的汇总，再也不用在标签页之间来回横跳。重复性的浏览器操作也能一句话交给它跑，我只管看结果、做决定。Chromium 内核，常用插件无缝迁移，还比 Chrome 轻快不少。已经回不去了。#TabbitBrowser

[qinbafrank @qinbafrank]

今天国内的金融领域的会关注度很高，降息降准+降存量房贷利率和首付比例+结构性货币工具。8月中潘功胜接受央视采访提到的三个要点这次都应验了。美联储的降息给了国内货币金融政策更大的操作空间，前两项都属于加大货币政策实施力度的范畴，第三项结构性货币工具应该就是新的增量政策。细说下对这个结构性货币政策工具的理解

央行创设结构性工具面向两类主体投放流动性：
一是面向非银类机构，抵押风险资产获得无风险资产，去进行市场化融资，资金可用于购买股票；
二是向银行发放再贷款用于贷款给上市企业进行股票回购或增持。

两类工具的出发点都是为了打通资金进入权益市场的渠道，但跟很多人的理解不一样：非银机构拿着手里的股票ETF、沪深300成份股等资产抵押从央行手里换国债和央行票据，本质还是资产的换手流动性的调配。非银机构还要把这些国债和央行票据卖掉才能换回资金，所以并没有新增资金进来，只是让非银机构把资金从国债市场往股市来抽。

当然看起来还有个意图就是打压近期高涨的国债购买热情，毕竟十年国债收益率都快到2%了。所以央行���没有印钞放水，只是在不同资产间调配流动性。至于再贷款给上市企业用户回购、那就更没有放水、最多是降低了贷款。成本。

与央行直接购买的区别在于，后续见效程度取决于市场主体进行此类操作的意愿与动力。短期应该有一波行情，毕竟刺激了情绪和预期（可能未来进一步开口子），中期除了这些非银机构和企业低息贷款回购股票的意愿外、那就看基本面能不能改善或者改善的程度了。

[Viking @vikingmute]

昨天看到这个开源库挺棒：Sidekick
https://t.co/ZPZ2QwPruu

是使用 VPS 时候配合非常好的 Cli 工具，可以很轻松的做一键配置/部署以及集成 CI/CD 还有 HTTPS，有点像之前 DHH 他们出品的 Kamal。功能包括：

* 一键配置 VPS 环境（创建用户/安装 Docker/Traefik等等）
* 一键部署，本地一个 Dockerfile 就够
* Zero downtime
* 自带 load balancing
* 零配置 SSL Certs

我用 DO 的一台机器试了一下，确实简单好用，有点像自建版本的 https://t.co/U6fheGFgPv

[piglei @Piglei]

《13 年后，我如何用 Go 写 HTTP 服务》

https://t.co/vu8jm4J7KB

一篇 Go 语言方面的最佳实践类��章，只涉及标准库中的 HTTP 基建，不涉及其他第三方 Web 框架或库。作者有十余年的 Go 编程经验，经验丰富。

文章除了展示具体的代码编写与组织技巧，也谈了一些“为什么如此处理”背后的设计考量，包括：长参数列表的函数、请求编解码处理、用闭包结合 http.Handler、E2E 测试和单元测试，等等。透过这些考量，能感受到作者多年经验与智慧的沉淀。

作为一个极少用 Go 写 HTTP 服务的人，我也读得津津有味。

[AIGCLINK @aigclink]

一个使用最小熵耦合技术的隐写工具：Tomato，可以将加密信息隐藏在自然语言文本中

可以用于信息安全、数据保护、隐蔽信息传递、数字水印、艺术创作等领域

特点：
1、使用LLM生成掩护文本，通过最小熵耦合技术将加密信息嵌入到掩护文本中，确保了嵌入的信息在统计特性上与自然语言保持一致，难以被检测到

2、在嵌入时，通过合并掩护文本和隐藏信息的概率分布来最小化联合熵

3、在解码时，利用LLM的上下文感知能力，使用MEC反向解耦提取隐藏信息

4、可定制，比如加密密钥长度、文本长度、模���选择等，可以根据需求调整

github：https://t.co/iS1rdL6Oie

#Tomato #LLM #Steganography #DataSecurity

[The Rust Dev @the_Rust_Dev]

Dive into Rust's ownership model. Discover how it manages memory with stack and heap, ensuring safety and efficiency. Great read for devs!

{ author: @cudidotdev } #DEVCommunity #RustLang
https://t.co/sStJ50Ya2B

[The Go Dev @TheGoDev]

Learn how to build a music streaming service with Python, Golang, and React, focusing on system design and anti-piracy measures.

{ author: @koladev32 } #DEVCommunity #GoLang
https://t.co/LMr4luYKrg

[Golang News & Libs & Jobs - human 🗣️ , no 🤖 @golangch]

"Time to Leave? Time to Rebuild! Making Twitter2.0 with Go" by @lukeocodes
#golang #DEVCommunity

https://t.co/R2eRrxet2z

[elvis @omarsar0]

Top most interesting AI papers I read this week:

1/ SAM 2: An open unified model for real-time, promptable object segmentation in images and videos. SAM 2 can be applied to unseen visual content without the need for custom adaptation. To enable accurate mask prediction in videos, a memory mechanism is introduced to store information on the object and previous interactions. The memory module also allows real-time processing of arbitrarily long videos. SAM2 significantly outperforms previous approaches on interactive video segmentation across 17 zero-shot video datasets while requiring three times fewer human-in-the-loop interactions. https://t.co/ZAhTdhsi5b

2/ Structured Generation Limits Reasoning: Investigates if structured generation can impact an LLM’s reasoning and domain knowledge comprehensive capabilities. Observes that there is a significant decline in LLM’s reasoning abilities when applying format restrictions compared to free-form responses. This degradation effect is further amplified when applying stricter format constraints to reasoning tasks. https://t.co/45UFbSkLpw

3/ From LLMs to LLM-based Agents for Sofware Engineering: A survey paper on current practices and solutions for LLM-based agents for software engineering. Covers important topics such as requirement engineering, code generation, test generation, and autonomous decision making. It also includes benchmarks, metrics, and models used in different software engineering applications. https://t.co/O6QQTK9Z9I

4/ Enhancing LLMs for RAG - Introduces RAGFoundry, an open-source framework for augmented LLMs for RAG use cases. It supports data creation, training, inference, and evaluation. One useful application is the creation of data-augmented datasets for tuning and evaluating LLMs in RAG settings. https://t.co/kqoCkcrJch

5/ Self-Taught Evaluators: An approach to improve model-based evaluators using synthetic training data only. It first generates contrasting outputs (good and bad model responses) and trains an LLM-as-a-Judge to produce reasoning traces and final judgments. The self-improvement scheme repeats the training process in an iterative way using its improved predictions. Claims to outperform LLM-judges such as GPT-4 and match top-performing reward models trained on labeled examples. It improves a strong LLM (Llama3-70BInstruct) from 75.4 to 88.3 (88.7 with majority vote) on RewardBench. https://t.co/GIPjfyrAId

There are a few more exciting papers that I will highlight tomorrow in the Top ML Papers of the Week @dair_ai. Stay tuned!

[ClickHouse @ClickHouseDB]

At ClickHouse, super-fast aggregations are our thing. But how does it work under the hood? 🥷

By parallelizing absolutely EVERYTHING! 🚀🚀🚀 and in today’s blog, we explore those parallelization techniques in depth 🤓

https://t.co/1Ccb0e9Bms

[宝玉 @dotey]

我们编程界的至理名言：Make it work, make it right, make it fast

[Ivan Fioravanti @ivanfioravanti]

1/7 Fine tuning in Apple MLX, GGUF conversion and inference in @ollama ? Yes you can! 🔥🔥🔥
Here the steps to follow:
- convert a model from hf
- fine tune
- fuse
- llama.cpp https://t.co/RLw8jCP4mS & quantize
- ollama modelfile & create
- you are ready to go! HAVE FUN!

[Garfield | 𝐦𝐞𝐦𝐨𝐫𝐲_𝐨𝐫𝐝𝐞𝐫_𝐚𝐜𝐪_𝐫𝐞𝐥 @Garfield9888]

现在是rc-table contributor

[Donald J. Trump @realDonaldTrump]

Thank you Helsinki, Finland!

[NSF - NASASpaceflight.com @NASASpaceflight]

ARTICLE: Parker Solar Probe, ULA enter final pre-launch processing marathon -

https://t.co/VFVKWfTf3n

- By Chris Gebhardt (@ChrisG_NSF)

[Garfield | 𝐦𝐞𝐦𝐨𝐫𝐲_𝐨𝐫𝐝𝐞𝐫_𝐚𝐜𝐪_𝐫𝐞𝐥 @Garfield9888]

@VirginAustralia Boeing 737-800 and any prizes?