ทวีตที่ปักหมุด
💜🥰💕💙 my first official release in ages 🩵💛❤️💕
A remix for my good friend Zphr. Pls enjoy ☺️
open.spotify.com/track/31cUz0ZK…
English
madison
11.3K posts
madison
@_madison______
welcome to my twitter page
Philly เข้าร่วม Şubat 2009
1.3K กำลังติดตาม666 ผู้ติดตาม
@teej_dv @dillon_mulroy Do you reach for alchemy over sst in general these days? I’m overdue to give alchemy a go and would like to hear the experience of folks who have tried both.
English
@dillon_mulroy have a similar thing and it works great (altho have it set up to do alchemy doing the deploys when someething changes via actions)
English
migrating all my personal apps and agents on cloudflare to a single monorepo has been a clutch move
English
@GoodmanHoops the pbp says there were only 2 missed shots in the last 4:30 haha
English
That was one of the best conference tournament games I've EVER seen.
High, high level.
Arizona wins 82-80 over Iowa State on the Jaden Bradley buzzer-beater.
Dell'Orso and Kharchenkov were sensational, Momcilovic hit 8 's for Iowa State.
Bradley was the guy in crunch time. Again.
English
@irl_danB @JakeBlockchain Would love to see an example of the implementation of the last one if you’re willing to share
English
it’s mostly a lot of DevOps stuff
many kicked off by GitHub actions to do reviews, security scans, etc.
a lot used for testing: spawning multiple test users to interact with my API while others monitor logs/database, all in parallel
doing a lot of codebase-mental model sync stuff, one of my favorites is one that quizzes me about the current state of the codebase while a fan-out crawls it and checks for divergence between my mental model and code on the ground, then offers to bridge the gaps either by teaching me or patching the code
English
greetings Dario
I canceled my ChatGPT Pro sub in favor of Claude Max for the last several months. big fan, very loyal customer
however you probably now have somewhere between a day and a week to introduce 1000tps to Claude Max or you’ll lose me to the dark side again
Kevin Weil 🇺🇸@kevinweil
Coding at 1000 tokens/sec is a mind-expanding experience. You have to try this.
English
@ScriptedAlchemy is there currently any tool you're aware of besides the cursor long-running agents you've been using that can do that reliably?
English
I should be able to one-shot a million line pr and it be mostly correct. That’s now table stakes for AI products. Don’t yield back to the user unless the task is done or you need help. I don’t want to say “continue” 100 times; it’s ludicrous that we ever had to do that at all.
English
@natebirdman I need help here lol. As someone who generally used to skip the design step and just “write the code till it look good”, I haven’t been able to move to a fully agent workflow yet there. “From the perspective of an award-winning mobile designer” ain’t cutting it.
English
@irl_danB They say its the base model at 0:42 here x.com/ryanvogel/stat…
vogel@ryanvogel
introducing a new weekly series focused on @opencode oc weekly: episode 1 (volume up)
English
@irl_danB Sounds like Opencode Zen has it models.dev/?search=arcee
English
@cognition @cursor_ai @opencode @vercel @julesagent @AmpCode @Cloudflare @savarlamov wtf is this music
English
Agent Trace: Capturing the Context Graph of Code
We are delighted to collaborate with @cursor_ai, @opencode, @vercel, @julesagent, @ampcode, @cloudflare, and @savarlamov in an open standard for mapping back code:context.
here's how we see the potential of code context graphs and the new era of better tooling and better agents it enables.
(yes the following is vibe-videoed with @remotion's Skill and @windsurf, 100% ai edits incl audio)
English
@Grady_Booch Remember when that guy asked you if you had ever heard of uml?
English
Excited to launch happycapy, an agent-native computer on your browser.
Claude Code → Clawdbot → Happycapy
> Private sandbox for everyone to run Claude Code anytime anywhere - no Mac Mini needed
> GUI built for everyday user, visualizing skills and its output
English
Should we do more explicit launch podcasts like this?
(it was a lot of fun)
Nathan Lambert@natolambert
Here's my conversation with @latkins and the team at @arcee_ai on their path to training and releasing Trinity Large today. From going all in on open models built end to end in the US 6 months ago to having the model in hand is no easy feet. I loved this conversation on how to design a startup around open models and take a bold step to scale it up. I'm openly an Arcee fan, watching them take risk and pull it off. We discuss: - The state (and future) of open vs. closed models, - The business of selling open models for on-prem deployments, - The story of Arcee AI & going “all-in” on this training run, - The ATOM project, - Building frontier model training teams in 6 months, - and other great topics. I really loved this one, and think you well too. Chapters: 00:00:00 Intro: Arcee AI, Trinity Models & Trinity Large 00:08:26 Transitioning a Company to Pre-training 00:13:00 Technical Decisions: Muon and MoE 00:18:41 Scaling and MoE Training Pain 00:23:14 Post-training and RL Strategies 00:28:09 Team Structure and Data Scaling 00:31:31 The Trinity Manifesto: US Open Weights 00:42:31 Specialized Models and Distillation 00:47:12 Infrastructure and Hosting 400B 00:50:53 Open Source as a Business Moat 00:56:31 Predictions: Best Model in 2026 01:02:29 Lightning Round & Conclusions More great open model builder podcasts coming soon!
English
It’s crazy how much this smarter vs faster spectrum impacts dev workflow.
I switched from opus to 5.2 because I got hooked on “smarter” and it’s changed my work day completely.
Instead of needing 3 hour blocks of time to get in the zone and accomplish something, now I want more 30 min blocks in the day to respond to async agents and keep them going
Andrej Karpathy@karpathy
@Rasmic I hope this is not my fault. It's definitely very smart so a little bit faster would be good now. x.com/karpathy/statu…
English
@elder_plinius This is objectively the most optimal use for this gif
English
English
Our AI bot hallucinated that the Web UI has a logging UI. But I am totally going to build one because I see how useful this will be to help new people.
HDD: Halucination Driven Development?
English
@LukeParkerDev actually, I put that part in the AGENTS.md. The plan was informed by my initial attempt to monitor network requests and build a hacky API client, so I first had the model try that. It learned a lot about the task, and used that in the plan creation (same session)
English
@LukeParkerDev telling it "use playwright mcp to validate your approach along the way. dont guess" and that seems to have been enough.
English
@LukeParkerDev hey I used your ralph TUI to build this agent-focused scraper CLI with a ton of bells-and-whistles for insurance research. 1-shot over 3 hours. Love the TUI experience. Thanks for throwing that together.
github.com/madisonbullard…
opncd.ai/share/ANk1SJm5
English