cookiefabricator

10.8K posts

cookiefabricator banner
cookiefabricator

cookiefabricator

@xaviorhornindal

Katılım Eylül 2012
826 Takip Edilen286 Takipçiler
Audie Murph
Audie Murph@bodysincloset·
Professional development tip of the day: it takes 5 minutes to set up a new gmail
Audie Murph tweet media
English
131
90
5.6K
480.9K
cookiefabricator
cookiefabricator@xaviorhornindal·
@ChaiDeluxe im not paid to know what im doing, im paid because i can google faster than everyone else
English
0
0
0
207
cookiefabricator
cookiefabricator@xaviorhornindal·
@Apple or @Steam should do the funny thing and give its devs a 15% discount if not ai art or code was used
English
0
0
0
24
cookiefabricator
cookiefabricator@xaviorhornindal·
@om_patel5 Unpopular opinion. AI should know. Tell us how to be better at optimizing its token usage. Or better yet have a top layer that prevents us from wasting its time converts all the text to be better for Anthropic.
English
0
0
1
171
Om Patel
Om Patel@om_patel5·
THIS GUY AUDITED 926 CLAUDE CODE SESSIONS AND FOUND MOST OF THE TOKEN WASTE WAS ON HIS SIDE everyone is blaming anthropic for the limits, so he decided to actually look at the data 858 sessions, 18,903 turns, and $1,619 estimated spend across 33 days here's what he found: 1\ one default setting was burning 14,000 tokens per turn Claude Code loads the full JSON schema for every tool into context at session start. whether you use them or not. 20,000 tokens of tool definitions sitting there on every single turn. the fix: one line in your settings.json "ENABLE_TOOL_SEARCH": "true" context dropped from 45K to 20K instantly. across 858 sessions that one setting was wasting an estimated 264 million tokens 2\ cache expiry is the single biggest waste 54% of his turns came after a 5+ minute idle gap. every one of those turns re-processed the entire conversation at full price which caused a 10x cost jump you go grab coffee. come back 5 minutes later. type your next message. everything rebuilds from scratch. the context didn't change. you didn't change. the cache just expired. 12.3 million tokens wasted on idle gaps alone 3\ 42 skills loaded. 19 of them used twice or less across 858 sessions. every one of those skill schemas sat in context on every turn eating tokens for nothing. 4\ 1,122 redundant file reads where the same file was read 3+ times one session read the same file 33 times. he ALSO built a full token auditor dashboard that shows you exactly where your waste is coming from 19 charts, opens in your browser, free AND open source
English
97
154
1.9K
285.3K
cookiefabricator
cookiefabricator@xaviorhornindal·
@miaaowing Considering I see people using caps lock instead of shift for capitalization. I don’t think anybody uses the delete key except for the higher ups.
English
0
0
0
26
cookiefabricator
cookiefabricator@xaviorhornindal·
@thsottiaux local compute options to help offset loads and extend token usage. like a sub agent
English
0
0
0
11
Tibo
Tibo@thsottiaux·
With Codex the there is quite the gulf in load between peak and off-peak times, and we would like to achieve more of a smoother traffic pattern as that would be a more optimal use of our compute. We have ideas, but curious what you all think we should do? Would more usage during off-peak and surge multiplier during peak times make sense?
English
795
43
1.7K
204.6K
cookiefabricator
cookiefabricator@xaviorhornindal·
@vishaltweetup unlimited online prompts and a decent amount of coding and a better visability engine when coding. yea its great
English
0
0
0
21
Vishal
Vishal@vishaltweetup·
Unpopular opinion: If you’re paying $20/month, ChatGPT Plus is better than Claude Pro.
Vishal tweet mediaVishal tweet media
English
318
35
2.4K
459.8K
cookiefabricator
cookiefabricator@xaviorhornindal·
@memecrashes I’ll take Google slides and Google sheets any day over Microsoft bullshit
English
0
0
15
3K
cookiefabricator retweetledi
•rottenangel
•rottenangel@MimisDiarys·
Face Card 10/10
English
54
148
13.2K
177.6K
cookiefabricator
cookiefabricator@xaviorhornindal·
@rohanvarma i actually do this with ollama, i have it compile a massive list of data then use something like claude to review and integrate
English
0
0
0
7
Rohan Varma
Rohan Varma@rohanvarma·
If we made /slow mode in Codex, would you use it? What for? (Slower inference at a cheaper cost)
English
952
32
2.2K
185.7K
Atharva
Atharva@AtharvaXDevs·
NEVER BUY A CLAUDE CODE 20$ PLAN IT’S JUST A SCAM😭 NEVER BUY A CLAUDE CODE 20$ PLAN IT’S JUST A SCAM😭 NEVER BUY A CLAUDE CODE 20$ PLAN IT’S JUST A SCAM😭 NEVER BUY A CLAUDE CODE 20$ PLAN IT’S JUST A SCAM😭 NEVER BUY A CLAUDE CODE 20$ PLAN IT’S JUST A SCAM😭
English
484
208
6.5K
1.3M
cookiefabricator retweetledi
Princess-Rae
Princess-Rae@theprincessrae_·
Happy BST 🌞
English
40
77
1.1K
51.9K
cookiefabricator
cookiefabricator@xaviorhornindal·
@snwy_me Honestly, Lennox is fine enough on the server side. I just need apple to join the enterprise active directory side of things and managing stuff from a fleet of MacBooks.
English
0
0
0
20
snwy
snwy@snwy_me·
apple had a server line (XServe) but discontinued it because they had no advantage over other servers. nobody bought them with Apple Silicon there's a real reason for Apple servers to exist - imagine an M3 Ultra but designed without desktop thermal/power constraints in mind
snwy tweet media
snwy@snwy_me

apple *has* to have a successor lined up, given how lucrative Apple Silicon is for ML it would be bad business to not have a real Dell GB300 tower competitor, right? they also really need to start getting these in actual racks

English
81
60
1.8K
139.4K
cookiefabricator
cookiefabricator@xaviorhornindal·
@ThePrimeagen always a struggle to get signed into the app, i joke that it must be a vibe coding issue lol
English
0
0
1
31
ThePrimeagen
ThePrimeagen@ThePrimeagen·
Claude is too dangerous
ThePrimeagen tweet media
English
89
25
1.4K
90.1K
The Lunduke Journal
The Lunduke Journal@LundukeJournal·
CachyOS, an Arch-based Linux distro focused on performance has: - Taken the top spot on the DistroWatch rankings. - Become the most used Linux distribution on ProtonDB (a website for tracking game compatibility on Linux). CachyOS has also hit 8.5% of all Linux users of Steam (beating out Ubuntu, Mint, & Bazzite).
The Lunduke Journal tweet mediaThe Lunduke Journal tweet mediaThe Lunduke Journal tweet media
English
73
50
895
50.2K
Shishir
Shishir@ShishirShelke1·
Someone added proper cooling to the MacBook Neo 💀 Using a custom copper heatsink over the CPU, then attaching a liquid-cooled Peltier cooler externally And the results are insane: • Temps: 105°C → ~74°C • Gaming: ~30fps → ~58-60fps • Benchmarks: +17-18% Sustained performance improved massively with no aggressive throttling! 📸 via: ETA Prime / YouTube
Shishir tweet mediaShishir tweet mediaShishir tweet mediaShishir tweet media
English
55
145
3.9K
290.7K
cookiefabricator
cookiefabricator@xaviorhornindal·
@OutofGalaxyy Gets better for gaming laptops since some have better thermal paste or better powerbanks for extra performance. Like 10-20% difference with same specs lol
English
0
0
0
36
Out of Galaxy
Out of Galaxy@OutofGalaxyy·
While searching for a new Windows Laptop, I’ve realised one thing: the chip naming system on current Windows computers is SO FUCKING BAD Cause wdym you have Intel with their Core i3, i5, i7, i9, Core 3, 5, 7, 9, Core Ultra 3, 5, 7, 9, then you have versions like 256U, 255U, 235U, 225U, models like 14900KS, 14900K, 14900, 14900KF, 14900F, 14700, 14700T, 14600K etc etc etc Then you have AMD with their Ryzen 3, 5, 7, 9 2000 series, 3000 series, 4000 series, 5000 series, 7000 series, 8000 series, 9000 series, Ryzen AI 5 440G, 440GE, 435G, 435GE, AI 7 450G, 450GE etc etc Like HOW AM I SUPPOSED TO KNOW WHICH ONE TO GET A LAPTOP WITH??? How am I supposed to understand all that???? Apple has an M3 Ultra, M4/Pro/Max, M5/Pro/Max, and an A18 Pro in their MacBook Neo available right now. THATS IT.
English
294
91
2.8K
228.4K
Ramin Nasibov
Ramin Nasibov@RaminNasibov·
I wanna know what idiot at HP thought THIS would be a great place to put a power button??
Ramin Nasibov tweet media
English
952
868
22.9K
956.3K