Sabitlenmiş Tweet
Revolve
18.8K posts


@zerohedge How meta is not bankrupt as they are lagging in every front
English
Revolve retweetledi
Revolve retweetledi
Revolve retweetledi
Revolve retweetledi
Revolve retweetledi
Revolve retweetledi

In 72 hours I got over 100k of value
1. Lambda gave me 5000$ credits in compute
2. Nvidia offered me 8x H100s on the cloud (20$/h) idk for how long but assuming 2 weeks that'd be 5000$~
3. TNG technology offered me 2 weeks of B200s which is something like 12000$ in compute
4. A kind person offered me 100k in GCP credits (enough to train a 27B if you do it right)
5. Framework offered to mail me a desktop computer
6. We got 14,000$ in donations which will go to buying 2x RTX Pro 6000s (bringing me up to 384GB VRAM)
7. I got over 6M impressions which based on my RPM would be 1500$ over my 500$~ usual per pay period
8. I have gained 17,000~ followers, over doubling my follower count
9. 17 subscribers on X + 700 on youtube.
The total value of all this approaches at minimum 50,000$~ and closer to 150,000$ if I leverage it all.
---------------------
What I'll be doing with all this:
Eric is an incredibly driven researcher I have been bouncing ideas off of over the last month.
Him and I have been tackling the idea of getting massive models to fit on relatively cheap memory.
The idea is taking advantage of different forms of memory, in combination with expert saliency scoring, to offload specific expert groupings to different memory tiers.
For the MoEs I've tested over my entire AI session history about 37.5% of the model is responsible for 95% of token routing.
So we can offload 62.5% of an LLM onto SSD/NVMe/CPU/Cheap VRAM this should theoretically result in minimal latency added if we can select the right experts.
We can combine this with paged swapping to further accelerate the prompt processing, if done right we are looking at very very decent performance for massive unquantisation & unpruned LLMs.
You can get DeepSeek-v3.2-speciale at full intelligence with decent tokens/s as long as you have enough vram to host the core 20-40% of the model and enough ram or SSD to host the rest.
Add quantisation to the mix and you can basically have decent speeds and intelligence with just 5-10% of the model's size in vram (+ you need some for context)
The funds will be used to push this to it's limits.
-----------------
There's also tons of research that you can quantise a model drastically, then distill from the original BF16 or make a LoRA to align it back to the original mostly.
This will be added to the pipeline too.
------------------
All this will be built out here: github.com/0xSero/moe-com… you will be able to take any MoE and shove it in here, and with only 24GB and enough RAM/NVMe to compress it down. it'll be slow as hell but it will work with little tinkering.
------------------
Lastly I will be looking into either a full training run from scratch -> or just post-training on an open AMERICAN base model
- a research model
- an openclaw/nanoclaw/hermes model
- a browser-use model
To prove that this can be done.
--------------------
I will be bad at all of it, and doubt I will get beyond the best small models from 6 months ago, but I want to prove it's no boogeyman impossible task to everyone who says otherwise.
--------------------
By the end of the year:
1. I will have 1 model I trained in some capacity be on the top 5 at either pinchbench, browseruse, or research.
2. My github will have a master repo which combines all my work into reusable generalised scripts to help you do that same.
3. The largest public comparative dataset for all MoE quantisations, prunes, benchmarks, costs, hardware requirements.
--------------------------
A lot of this will be lead by Eric, who I will tag in the next post.
I want to say thank you to everyone who has supported me, I have gotten a lot of comments stating:
1. I'm crazy, stupid, or both
2. I'm wasting my time, no one cares about this
3. This is not a real issue
I believe the amount of interest and support I've received says it all.
donate.sybilsolutions.ai

English
Revolve retweetledi
Revolve retweetledi
Revolve retweetledi
Revolve retweetledi

I tested something I'd written myself: my first academic article, published 45 years ago. It came in at 77% AI-generated.


Prof. Devi Sridhar@devisridhar
Exact same issue for me- I know my previous books and articles have been used to train AI (looking at you anthropic)- & when I run previous articles (written pre-AI) into AI checkers, they can come back as high as 90% AI. It's not artificial intelligence- it's collective human intelligence.
English
Revolve retweetledi

BREAKING: To those asking what is burning behind the mountains north of Damavand, east of Tehran: it is an Iranian MODAFL complex where nuclear warheads for intercontinental ballistic missiles were designed, developed, and produced—intended for use against targets such as London. That complex has now been completely destroyed. U.S. Air Force B-1B heavy bombers carried out the strike.
#OperationEpicFury #OperationLionsRoar

English
Revolve retweetledi
Revolve retweetledi

BREAKING: This new footage shows the moment an ammunition and weapons depot at the 4th Tactical Fighter Base (Vahdati Airbase) of the Iranian Air Force in Dezful, southwest Iran, exploded after being hit by BGM-109 Tomahawk cruise missiles launched by the US Navy today.
#OperationEpicFury #OperationLionsRoar
English
Revolve retweetledi

Chuck Norris held a 183-10-2 record and was a 6x world champion in full contact bare knuckle karate.
On top of that, he beat heavyweight kickboxing world champion Joe Lewis 3 consecutive times and also had a brutal sparring match with undefeated kickboxing world champion, Bill Superfoot Wallace, that lasted an hour and a half. According to Wallace, they practically stalemated and "beat the crap out of each other".
Chuck was trained in kickboxing/boxing by Benny The Jet Urquidez and was also trained in BJJ by the Gracies and Machados for 20 years. Even being able to submit Carlos Machado himself on occasion.
Chuck had a 315 Ibs bench press at 180 lbs bodyweight and was said to have a grip back in the day that nobody could escape from because he was so strong. Even Jean Claude Van Damme said he'd never fight Chuck Norris, despite being a kickboxing world champion himself.
Chuck held a 10th degree black belt in Chun Kuk Do, a 9th degree black belt in Tang Soo Do, an 8th degree black belt in Taekwondo, a 5th degree black belt in Karate, a 3rd degree black belt in Brazilian Jiu Jitsu and a black belt in Judo.
Rest in peace, Chuck!
English
Revolve retweetledi

Qatar Dethroned As 'LNG King' As U.S. Seizes Throne, Reshaping Future Of Gas zerohedge.com/energy/qatar-d…
English
Revolve retweetledi














