rideroute.xyz

The Real House Perp Experiment. (allegedly) (supposedly)
🚂 Come aboard: BTC & ETH, 1,200×.

[rideroute.xyz @rideroute_trade]

wagwan chooms 🚂
me and a few others have been building Route—house perp on HyperEVM.

the idea: look at what our bloodline @papertrade_xyz and @cardboard_trade built, and see what we could add.
some say 3rd place is last. we say it's just different.

testnet's live, check bio~

[rideroute.xyz @rideroute_trade]

@jonathan_wilke more like the api is considerably marked up compared to the plans cuz most api use is enterprise

[rideroute.xyz @rideroute_trade]

这就跟劝你干啥都换左手、别用右手，免得左手的精细活儿退化一样

谁在乎啊

[马天翼 @fkysly]

建议大家千万不要用 Claude，千万不要依赖。

多用用 GPT Sol、Gemini、国模之类的。

否则的话，你会发现你只需要说一句话就能干完活了，养成了依赖性。

用 Claude 会使得你缺少了：
- 反复优化 Prompt
- 反复验证
- 写详细方案
- Review 每个细节
的这些麻烦事情。

说白了，用太好用的模型，你的 AI 能力将会大幅度下降，用进废退。而多用差点的模型，可以有效训练你的把控能力。当你哪天再遇到差模型的时候，你才会知道该怎么做。

[rideroute.xyz @rideroute_trade]

@blknoiz06 exactly what would be built on it tho, if u could have an idea i guess of sorts, u mean js another front end for the same underlying platform, or something actually meaningful

[rideroute.xyz @rideroute_trade]

it might be a bit hard to time, but generally speaking, in the middle of a rather reasonably large trend happening, u can make good consistent profit in a house perp if ur witty with leverage

and ay while 1000x is risky, if u happen to be blessed by the gods of trading and timed a eth short the moment news came out bout huang resigning from nividia, u'd have made quite the fat big bag

(depending on ur margin)

the concept's sexy; the execution is looking butter-faced on the big players side

[Ansem 🐂🀄️ @blknoiz06]

am really liking @papertrade_xyz, the high leverage individual positions w/ no worry about order book fills is very good for low timeframe scalping, especially in a trending market

think ppl are still underappreciating this as a protocol & $PAPER stakers are still printing cash off rewards

[rideroute.xyz @rideroute_trade]

i don't think anyone underappreciated the protocol, i think most ppl felt cheated by promises made vs promises kept my dude

it's a really exciting concept tho, hence why we even made our own version, but like what do they call it?
rocket in bottle or sum?? no it was lightning

they had lightning inna bottle and decided to toss it into the bin

[rideroute.xyz @rideroute_trade]

@DegenApe99 😭holy fuck choom is it so hard to maintain a platform????? they act like its giving them grief to maintain these things which don't really require as much maintenance as u'd think

i swear this is looking bad for the ecosystem as a whole

[rideroute.xyz @rideroute_trade]

@WatcherGuru this is sounding more and more like a deliberate attack on Huang and potentially local models.....cuz fym all our models stolen? him fucking a Chinese spy doesn't mean he js rolled the red carpets.

bro who tf is benefiting from these press releases, cuz someone clearly is

[rideroute.xyz @rideroute_trade]

good, wtvr continues the betterment of open models i stan, may they get all the gpu's they could ever want in the world, as long as they keep releasing said models for local, ablated use.

cool research on ur end tho chooms

[will depue @willdepue]

I have a major result to share.

I’ve found conclusive forensic evidence that the top 4 Chinese AI labs—DeepSeek, Moonshot, MiniMax and Ant Group—have built their newest models on NVIDIA’s most advanced Blackwell chips, which are barred from export to China.

This sounds insane, but I discovered that almost every AI model carries a hidden signature of the GPU that trained it. When a model is first set up, the starting values of its weights are laid out in a pattern that depends on the exact GPU. Each combination of GPU and seed number produces a unique pattern—a sort of fingerprint of the hardware used to create the model. Traces of that pattern survive training into the published model.

Again, this evidence isn’t based on any insider information or leaking. It’s based on a forensic analysis of the model weights released by the companies themselves.

I searched more than six trillion combinations of chip layout and seed number and matched them to open-weights models. In doing so, I found conclusive evidence that six Chinese AI models—including DeepSeek-V4-Pro, Kimi K3, and MiniMax-M3—were initialized on NVIDIA Blackwell systems. Four of the models match Grace Blackwell, the GB200 and GB300 parts in NVIDIA’s flagship liquid-cooled rack-scale systems.

I then verified these results on real hardware. Running unmodified PyTorch with the seeds I’d recovered, a Blackwell B300 reproduced DeepSeek-V4-Flash’s untouched starting weights bit for bit, and a rented GB200 reproduced Kimi K3’s to a correlation of 0.999995. A Hopper H100 reproduced none of DeepSeek-V4-Flash’s, and changing the seed by one broke the match. (DeepSeek’s seed is 953049583 by the way.) The method also was verified to match 24 known open-weights models to the hardware that developers said they used.

It is exceedingly unlikely that these results are a coincidence. Across six trillion combinations, the best accidental match should reach only about 8 sigma (i.e., eight standard deviations above what random chance would produce). The real matches range from 49 to 1,385 sigma, and even the weakest single test hits 20.

These were data-center-scale training runs. With some analysis of release schedules and model details, I estimate that the largest, Kimi K3 and DeepSeek-V4-Pro, used 2,000 to 3,000 Blackwell GPUs each. Clusters of this size cost on the order of $150 to $200 million each to buy, and across these labs alone would add up to more than half a billion dollars of export-controlled GPUs, not counting experimental compute or otherwise. It’s quite likely that we see multi-billion dollar expenditure on export-controlled GPUs.

I am highly confident in this result, and it has been independently verified by senior researchers at OpenAI and Anthropic. You can read the full report at https://t.co/YOtzND0Jqk.

There is much more I’ve identified about these models through other forms of checkpoint forensics like this, and will have more to share soon.

[rideroute.xyz @rideroute_trade]

@0xKento_ ts how u ended up wit claude deleting ur operating system cuz it was slowing you down

[rideroute.xyz @rideroute_trade]

@simpelyfe ts the funniest thing i seen pop up in a min

[rideroute.xyz @rideroute_trade]

@019ec6e2 im tryna be like bro tf

[rideroute.xyz @rideroute_trade]

it makes no sense to me as to why the ever fuck they would fire this guy. like jeez fucking louise gonk, he cheated on his wife with 2 women. Is it shitty that 1 happened to be a spy? yes. But are u gonna fucking oust him on the basis of what? moral integrity in his personal life? who gives af?

this choom is the leading edge for our ai and compute industries, and as long as he is fucking useful in any minute fucking sense he should never be discarded.

esp not for tim fucking cook bro like wha

[rideroute.xyz @rideroute_trade]

i have no clue what u do, however, if they gotta distill, then distill. only thing i personally care about is these models getting good enough that by the time claude gets fully enshittified and caged with a million safety parameters we have a great alternative that we can abilerate and run locally. So yes DISTILL THE FUCK AWAY CHINA, do it for the vine. Cuz i have no care about them devloping their own models, not a single ounce of care, they can do allat after claude gets enshittified. but thats js my personal op.

[rideroute.xyz @rideroute_trade]

@ihzn7 @jito_labs @cardboard_trade i'm lost, i thought cardboardtrade had no association with jito per their own reply? how does he even have access to the lp pool when they have a multi-sig for any changes??? wtf is happening

@cardboard_trade

[rideroute.xyz @rideroute_trade]

i guess this only works on mobile cuz ts aint work on my comp

deadass thought i got engagement farmed until i saw other ppls responses

[Lawrence Kitema @lawrencekitema]

I've been using X for 14 years, and I'm only just discovering that you can take a screenshot of a tweet directly from the Copy Link option. 😂

[rideroute.xyz @rideroute_trade]

@yenwod_ i mean to be fair choom,
u kinda did imply suggestively that anyone can learn hft

as for making money, thats an assumption that retail would make after reading what you said, if they deem it to be simple enough to enact (which they would cuz u did simplify it innit).

[rideroute.xyz @rideroute_trade]

@itsdistribution @voidfreud idk, i feel like its a zilch thing? i guess to each their own

open models exist that can be tuned, might not be the level of current claude but they get said job done to a degree

[rideroute.xyz @rideroute_trade]

@Stupifff no fucking way 💀

thats beyond me icl, i cant see how anyone would deem it profitable to spend 5 bags randomly on shit coins with the hope of one of it being a lucky winner

what are the chances? like 0.0 sum percent?

[rideroute.xyz @rideroute_trade]

@CrazyShyyt @grok read said article on arxiv and summarize the key points. do choomies claims check out?