Zakarth

419 posts

Zakarth

Zakarth

@Zakarth

Katılım Temmuz 2007
209 Takip Edilen39 Takipçiler
Zakarth
Zakarth@Zakarth·
@rdolmedo_ But they’re all just forms of exposure to patterns and attention. Given the right fine tune knobs anyone can make any model do anything. With Talkie one thing going for it and fine tuning on new data is that it hasn’t had weight pollution. New patterns stick out more.
English
0
0
3
249
Ricardo Olmedo
Ricardo Olmedo@rdolmedo_·
@Zakarth Post-training is all about aligning models with downstream tasks. Here the question is, what capabilities does pre-training on the internet give you that are not recoverable with a little post-training?
English
1
0
6
1.4K
Ricardo Olmedo
Ricardo Olmedo@rdolmedo_·
We fine-tuned Alec Radford’s 1930 vintage LLM to solve SWE-bench issues. After just ‼️250‼️ training examples, the model solves its first issue, a simple patch to the xarray library. 🧵👇
Ricardo Olmedo tweet media
English
23
79
1.1K
210.1K
Zakarth
Zakarth@Zakarth·
Talkie proves that people expecting AGI from LLM are cooked. A child could figure this out.
Zakarth tweet media
English
0
0
0
34
Zakarth
Zakarth@Zakarth·
Claude Code: this is going to take about a week’s worth of effort *rigs it up in 15 minutes* Claude Code: All done
English
0
0
0
24
Michael Hla
Michael Hla@hla_michael·
I trained an LLM from scratch on pre-1900 text to see if it could come up with quantum mechanics and relativity. While the model is too small to do meaningful reasoning, it has glimpses of intuition. When given observations from past landmark experiments, the model can declare that “light is made up of definite quantities of energy” and even suggest that gravity and acceleration are locally equivalent. I’m releasing the dataset + models and leave this as an open problem to the research community. I also include what this project has taught me about intelligence in a mini essay linked below. 🧵(1/n)
English
117
261
2K
311.6K
Zakarth
Zakarth@Zakarth·
Claude code “leak” is silly, it had already been RE’d for weeks prior and the code extracted. Anyone could have just used Claude Code to rebuild Claude Code.
English
0
0
0
47
Zakarth retweetledi
William Shatner
William Shatner@WilliamShatner·
During the first airing of my Star Trek series where a kiss was objectionable; many southern stations pulled the episode & condemned the show. Using today’s vernacular it would absolutely be called“woke DEI crap”because it went against “norms” of society for its time. Not a lot seems to have changed.🤷🏼😑
English
2.1K
5.6K
40.9K
1.3M
Zakarth
Zakarth@Zakarth·
@sudoingX Been following for a second but question — why Hermes vs say OpenCode ?
English
1
0
6
2.7K
Sudo su
Sudo su@sudoingX·
the founder of openclaw joined the company that was founded to make AI open and now charges you per token. and is now telling you open models aren't there yet. i run qwen 3.5 27b on a single 3090. 50 tok/s. it writes code, handles tool calls, runs agent sessions for hours. the model built a full space shooter, 3,000+ lines, from a single prompt. i published the data. "open models aren't there yet" is what you say when your harness can't parse tool calls on local models and you blame the model instead of fixing the harness. i have the DMs. people switch from openclaw to hermes agent and their "broken" models suddenly work. pair a good model with a good harness like hermes agent where parsers are built per model. your data stays on your machine. no API key. 0 subscription. no one training their next model on your thinking. don't listen to someone with an OpenAI paycheck telling you open source can't do the job. install it. test it yourself. the receipts are on my timeline. he built a harness that couldn't handle local models and chose the API paycheck over fixing it. that should tell you everything.
Peter Steinberger 🦞@steipete

@sbaratelli @nvidia @openclaw most folks will want as much intelligence as possible, and open models aren't there yet.

English
262
400
5.3K
412.4K
Zakarth
Zakarth@Zakarth·
@sudoingX Here’s a wacky setup: running an nvidia 5080 laptop GPU with a Radeon 890M. Trying to figure out an optimal way to use both… went down the Vulkan path, it works but definitely not great speed. Seems like the hardware is cursed for compromise!
English
0
0
0
40
Sudo su
Sudo su@sudoingX·
i just became a mod of x/LocalLLaMA. if you're running local models on your own hardware and want in, the community is open. pinned and highlighted on my profile. approving members starting today. drop your setup below and i'll get you in. 3060, 3090, 4090, 5090, AMD, whatever you're running. all welcome. if you're hitting issues with hermes agent, llama.cpp, model selection, configs, i'm here. let's make local AI accessible for everyone.
Sudo su tweet media
Sudo su@sudoingX

let me get you started in local AI and bring you to the edge. if you have a GPU or thinking about diving into the local LLM rabbit hole, first thing you do before any setup is join x/LocalLLaMA. this is the community that will help you at every step. post your issue and we will direct you, debug with you, and save you hours of work. once you're in, follow these three: @TheAhmadOsman the oracle. this is where you consume the latest edges in infrastructure and AI. if something dropped you hear it from him first. his content alone will keep you ahead of most. @0xsero one man army when it comes to model compression, novel quantization research, new tools and tricks that make your local setup better. you will learn, experiment, and discover things you didn't know existed. @Teknium maker of Hermes Agent, the agent i use every day from @NousResearch. from Teknium you don't just stay at the frontier, you get your hands on the tools before everyone else. this is where things are headed. if you follow me follow these three and join the community. you will be ahead of most people in this space. if you run into wrong configs, stuck debugging hardware, or can't get a model to load, post there so we can help. get started with local AI now. not only understand the stack but own your cognition. don't pay openai fees on top of giving them your prompts, your research, and your most valuable thinking to be monitored and metered. buy a GPU and build your own token factory.

English
324
41
813
61.2K
Zakarth retweetledi
Paul Snively
Paul Snively@JustDeezGuy·
At this point, there is no question that a disquieting percentage of people using LLMs—including developers using coding agents—are exhibiting EXACTLY the same reaction as many people first exposed to ELIZA around 1964-1966. The scale is different; the mechanism, not at all.
octo@the_octobro

Just one more prompt bro this time itll be a one shot youll see bro this time will be different just one more prompt i swear bro im gonna use a different model bro this time its gonna be different just one more prompt bro

English
19
40
702
90.7K
Zakarth retweetledi
Markov
Markov@MarkovMagnifico·
how my codebase written entirely with claude code runs
English
699
3.2K
63.8K
4.4M
TracketPacer
TracketPacer@TracketPacer·
ok so how do i dress up this labubu like a network engineer
English
33
1
120
8K
Zakarth retweetledi
IT Guy
IT Guy@T3chFalcon·
VPN companies spent millions convincing you that a hacker on Public Wi-Fi is reading your bank details. ​Meanwhile, HTTPS killed that threat 10 years ago. ​You are paying a monthly subscription to fix a 2010 problem. 💀
English
246
157
2.3K
144.9K
Zakarth
Zakarth@Zakarth·
For anyone who was burned on treklegame.com I accidentally had today’s word capitalized in the wordlist which made it impossible to win. If your streak was broken you can use treklegame.com/repairstats.ht… to fix yourself back up
English
0
0
0
24
spencer
spencer@techspence·
Lets test the X algorithm. Only like this if you follow me. If you don't follow me, ONLY comment, don't like it. In either case, DON'T repost.
English
57
5
268
16.3K
solst/ICE of Astarte
solst/ICE of Astarte@IceSolst·
Is there a test suite for SAST to compare tool accuracy?
English
12
0
21
3.4K
Zakarth retweetledi
bubble boi
bubble boi@bubbleboi·
Software engineers have negated every advancement in transistor density, computer architecture, compilers, and computer networking over the past 30 years.
bubble boi tweet media
English
404
2K
32.6K
1M
Knowing Better
Knowing Better@KnowingBetterYT·
@greggnunziata I'm pretty sure Texas is the only state that in additional to the National Guard, has its own militia - the Texas Guard - that exists completely outside of the US military structure.
English
7
1
103
3.8K
Gregg Nunziata
Gregg Nunziata@greggnunziata·
Texas used to boast about its special commitment to its sovereignty, now it's eager to hand over its militia to the federal executive's whims
Greg Abbott@GregAbbott_TX

I fully authorized the President to call up 400 members of the Texas National Guard to ensure safety for federal officials. You can either fully enforce protection for federal employees or get out of the way and let Texas Guard do it. No Guard can match the training, skill, and expertise of the Texas National Guard. They defend our country with pride. America must also know that Texas still has thousands of National Guard assisting with the Border security.

English
97
603
5.9K
112.2K
Matt Johansen
Matt Johansen@mattjay·
npm be like package.json { bio: "but most of all, samy is my hero" }
English
5
6
68
6.3K
Zakarth retweetledi
ThePrimeagen
ThePrimeagen@ThePrimeagen·
we are so back
ThePrimeagen tweet media
English
365
1.9K
24K
1.1M

Keşfet

@rdolmedo_ @hla_michael @sudoingX @TracketPacer @techspence @elonmusk @BarackObama @taylorswift13