spike

317 posts

spike banner
spike

spike

@alert090579992

From China. I love the fundamentals, first-principles thinking, designing and building, and the craft of getting the details just right.

Katılım Ekim 2024
44 Takip Edilen25 Takipçiler
spike retweetledi
Silentum 🦁
Silentum 🦁@xKhalifan·
BE THAT GUY Obsess over becoming better. Read old books. Learn underrated skills. Practice different sports. Master various martial arts. Study mathematics. Speak multiple languages. Become unpredictable. The healthiest person in the room. Moneymaxx. Help those in need. Inspire people. Show what greatness really is. In the pursuit of your life’s main purpose, your energy should, in parallel, go toward upgrading your character. Not only will it speed up the process of achieving your goals, but it will also make you feel more fulfilled throughout the journey.
amrit@amritwt

i think about this dude all the time

English
3
25
378
9.9K
spike retweetledi
Fernando
Fernando@Franc0Fernand0·
If you are a developer who likes to learn by doing, these 7 legendary GitHub repositories are gold: 1. Build your own X (525k stars) Step-by-step guides for building your own database, git, docker, or text editor from scratch. github.com/codecrafters-i… 2. Public APIs (450k stars) A huge list of free APIs for your next side project. github.com/public-apis/pu… 3. Developer Roadmap (361k stars) Interactive roadmaps for backend, frontend, DevOps and more. github.com/kamranahmedse/… 4. Project-based learning (273k stars) There are tutorials that show you how to make real applications. github.com/practical-tuto… 5. Computer Science from OSSU (206k stars) A complete CS curriculum from free online courses. github.com/ossu/computer-… 6. Tech Interview Handbook (141k stars) Practical content for every phase of a tech interview, from application to negotiation. github.com/yangshun/tech-… 7.Awesome Machine Learning (73k stars) A curated list of machine learning frameworks and libraries, organized by language. github.com/josephmisiti/a… Which one are you using right now?
Fernando tweet media
English
1
22
149
7.6K
spike retweetledi
Tech with Mak
Tech with Mak@techNmak·
It is dangerously easy to build a neural network today without actually understanding how it works. We live in an era of 'import torch'. You can train a model in three lines of code, but the moment you need to debug a collapsing loss function or a vanishing gradient, syntax won't save you. You need first principles. I recently went through this notebook collection by Simon J.D. Prince, and it is the antidote to tutorial hell. Instead of just showing you the code, it forces you to visualize the mechanics: 1./ The Math => It builds the intuition for shallow networks and regions before adding complexity. 2./ The Optimization => It doesn't just use an optimizer; it compares Line Search, SGD, and Adam so you see why they behave differently. 3./ The Modern Stack => It connects the dots from basic backpropagation all the way to Self-Attention and Graph Neural Networks. Move from running code to engineering systems => this is a goldmine. Here's the resource: udlbook.github.io/udlbook/
Tech with Mak tweet media
English
2
34
243
8.6K
spike retweetledi
Digi
Digi@digiii·
say this in the mirror every morning
Digi tweet media
English
5
105
1K
14.5K
spike retweetledi
Zhengyao Jiang
Zhengyao Jiang@zhengyaojiang·
The first experimental evidence of recursive self-improvement (RSI). Autoresearching the autoresearch agent for eight days. The result beats the harness we hand-tuned for two years, on held-out benchmarks: 🧵(1/7)
English
121
477
4.3K
472.5K
spike retweetledi
Rosmine
Rosmine@rosmine·
I fixed why LLMs write so poorly, and I have a demo to prove it Announcing Distribution Fine Tuning (DFT): A post training step that fixes LLM writing Model outputs fooled pangram on 100% of test cases
Rosmine tweet media
English
127
159
3.3K
539.4K
spike retweetledi
Linas Beliūnas
Linas Beliūnas@linasbeliunas·
Instead of watching another Netflix series tonight, watch this 16-minute talk from a Jane Street AI engineer explaining how the hedge fund uses LLMs to make $30B+ every year. It’s the clearest, most practical look I’ve seen at how a top quant firm actually builds custom LLMs and agentic tools for developers, especially when public models fall short on something like OCaml. Whether you’ve never built AI tooling before or you’re already shipping agents, you’ll walk away with concrete ideas on making LLMs reliable in production. Bookmark it & watch the talk today - it might be the most valuable thing you learn about real-world AI engineering all month.
English
8
23
338
40.8K
spike retweetledi
Shri
Shri@ElevateTSpirit·
The black sheep has always got the highest of potential, but most are too sensitive and even traumatised to the point that their biggest obstacle, their biggest enemy is their own self and that is where a lot of them lose. In self sabotage, in trying to be moral, fighting with their own emotions and their own shadow while others far below their talent simply play the game and recieve the treasure just because they were not their own worst enemy.
Moral Philosophy@ML_Philosophy

The most awakened person in a family is usually isolated, silent, and alone.

English
13
136
1.4K
51.4K
spike
spike@alert090579992·
about anysearch
spike tweet media
English
0
0
0
11
spike retweetledi
signüll
signüll@signulll·
you should apply this to this principle to daily life. e.g. whenever i create anything, i rarely if start with the artifact. i start with a *feeling*. before i text someone, write on here, design, build, or even share a photo, i ask myself one question: what do i want this person to feel? my favorite recent example is when i made a spotify playlist for someone. before i started to make it, i wrote a little note that captured exactly what i hoped they’d feel while listening. i only sent them the note after they finished the playlist. you get this nice little feedback loop this way too. & before someone comes at me, obviously i didn’t pioneer this process, lots of other amazing creatives did. but this is now second nature to me, it’s internalized.. “what do i want someone else (esp a person i love) to feel?”
Mike Wang@zmwang

Design for the feeling, first. I watched some of the best professors in the world design transformational classes. That informed my worldview for experience design (after all classes are events). They always started with outcomes first (what they wanted to teach), not content. This applies to events. What do you want your community to feel/know? The activities are abundant and can be molded.

English
14
23
455
33.7K
spike
spike@alert090579992·
更重要的一个原因是, 我希望掌控我的 agent, 这也是我使用 pi 而非 codex, 极少离开电脑使用的原因. 我需要掌控感和理解
中文
0
0
0
7
spike
spike@alert090579992·
我觉得我总是太过注重成本, 导致我手中还有5张reset, 且最近的一张在三天后就过期, 可我还是不敢放开手用codex... 很多时候想着 medium 够用 , low 够用, terra够用 ,就没必要开high, 上 sol... 我并不认为这是在帮openai省钱.. 只是习惯使然
中文
1
0
0
52
spike retweetledi
alexandru
alexandru@pondorasti·
herdr is the neovim of agentic engineering. and their greatest innovation might be the sidebar. I expect a lot more apps, both TUIs and GUIs, to start looking like this.
alexandru tweet media
English
8
4
131
7.2K
spike
spike@alert090579992·
@lassejv just downald, and run pi, it's really having a better using experience than windows terminal, and i am thinking to use it as my first choice. ignore my poor english haha🫡
English
0
0
1
7
Lasse
Lasse@lassejv·
If you haven't noticed termy.sh had a major rebranding. Big things are coming to termy 👀
English
4
0
9
2.1K
spike retweetledi
Hunter Leath
Hunter Leath@jhleath·
i've come to the conclusion that AI-generated prose is actually irredemable, and we've banned it from external comms at Archil not only does is it very easy to spot, but it easily contaminates your thinking making it *really* hard to recapture any sense of authenticity (once AI outputs an outline of talking points, it's harder to come up with better ones yourself) we've been bad about this in the past, and so i'm doing a manual rewrite of the docs this week to give a more obviously human tone
Matt Silverlock 🐀@elithrar

just saw probably the worst (so far...) AI slop deck ahead of an ext. meeting today (they are presenting to us). it goes without saying: please do not do this.

English
1
9
97
9.5K
spike retweetledi
tuna🍣
tuna🍣@tunahorse21·
increasingly convinced this is the best interface for agentic work basically a modern Franklin Servan-Schreiber zui that is infinite and has hierarchy lets me pull up and put away items etc
tuna🍣 tweet media
tuna🍣@tunahorse21

html might be the best canvas for agents btw was going back & forth with sol detailed plan, test etc still misaligned finally crash out and tell it to make an html, I will sort and move the task items, map out the issues etc, just export the json woked gg

English
7
3
120
7.4K
spike
spike@alert090579992·
ChatGPT网页端很卡有人有感觉吗?? 我是真希望有个TUI端的
中文
0
0
0
19

Keşfet

@lassejv @elonmusk @BarackObama @taylorswift13 @cristiano @BillGates @NASA @nikifrancismediavine