Shoestring Lab (@Shoestring_Lab) header
Shoestring Lab (@Shoestring_Lab) profile picture

Founder of Peopoli. Autist. Occasionally funny. https://teapaw.org/t.co/CbtZzVq1C4 Small business builder.

Nevada
Joined February 2024
Photo by Shoestring Lab (@Shoestring_Lab)Photo by Shoestring Lab (@Shoestring_Lab)Photo by Shoestring Lab (@Shoestring_Lab)Photo by Shoestring Lab (@Shoestring_Lab)Photo by Shoestring Lab (@Shoestring_Lab)Photo by Shoestring Lab (@Shoestring_Lab)Photo by Shoestring Lab (@Shoestring_Lab)Photo by Shoestring Lab (@Shoestring_Lab)Photo by Shoestring Lab (@Shoestring_Lab)Photo by Shoestring Lab (@Shoestring_Lab)
They're doing this because they got caught labeling black murderers as White.
Wake County, NC is deleting its entire online mugshot database Critics said they "fed stereotypes"
84
2,190
31
9,677
70,649
Truckee is buried. ❄️ Snow walls taller than people, sidewalks turned into tunnels, and cars half-swallowed. Locals shoveled this path this morning — the town’s overwhelmed and can’t keep up. Classic Sierra winter. #Truckee #Tahoe #Snow #SierraNevada #Winter #DonnerPass
83
236
32
1,573
59,287
Shoestring Lab retweeted
I watched this entire interview between Jeremy Boreing and Harmeet Dhillon, and what stuck out to me most was the constant admonishment of White and “Heritage” Americans for beginning to operate along ethnic lines and for seeking the betterment of their own ethnic in-group—that White Americans need to remain racially colorblind, since we’re all Americans, after all! The problem, however, is that this standard is seemingly only ever expected of American Whites. At this very moment on Capitol Hill, the Congressional Black Caucus (60+ members), the Congressional Jewish Caucus (20+ members), the Congressional Hispanic Caucus (50+ members), and the Congressional Asian Pacific American Caucus (24+ members) are all actively pushing legislation not based on what is best for ALL AMERICANS (because we’re all Americans, remember?), but rather legislation that will disproportionately benefit their own ethnic in-groups. What’s more, there are literally dozens of PACs dedicated to advancing the interests of every non-White American ethnic group—worth noting, much to the detriment of White Americans. These racially based PACs are collectively spending hundreds of millions of dollars every single election cycle on their own ethnically beneficial pursuits, I might add. If White Americans are expected to believe that we’re all in this together™, then perhaps that standard should be expected of every other ethnic group in the country as well—because it sure seems as if the all-seeing eye of inclusivity is exclusively focused on American Whites, while completely ignoring every other ethnic group.
"You cannot win elections in the United States with the white Christian vote. It does not math. It does not add up. You must have a coalition of people of all races, different beliefs, different origins." - Harmeet Dhillon @HarmeetKDhillon
205
746
52
3,946
62,283
Shoestring Lab retweeted
Over 800 Debian packages are impacted by compromised CA certificates. Steve McIntyre posted Debian scan results for compromised CA certificates, saying “And here are the results, compressed so they don't bounce off the message size limit. No security advisory has been published. mail-archive.com/debian-deve…
20
62
9
436
12,824
Shoestring Lab retweeted
The latest 0day is out... Sorry for dropping this on a Friday.
104
296
115
2,856
242,801
Shoestring Lab retweeted
TensorFold 1.3.0 is out 🚀 GLM-5.3-Flash went from 83.8 to 104.1 tok/s on two RTX PRO 6000 Blackwells. That’s about 24% faster on CUDA at TP=2 with NVFP4 experts, drafts off and the same output bits in that test. - On Metal, DeepSeek-V4.1’s prompt-attention kernel is about 1.5× faster, and GLM-5.3-Flash follow-ups reach the first token 3–4% sooner. - On CUDA, Flash Next frees about 3 GiB of page cache after loading the NVFP4 export. On Linux, you can opt in to mapping KV pages as streams need them. - tensorfold models now show which chat checkpoints fit your machine. On Metal, tensorfold info accounts for context and parallel streams when estimating memory for Qwen3.8-27B-class models. - Idol’s denoiser runs natively on Metal and on supported NVIDIA Blackwell GPUs. - The server has opt-in JSON request logs and a journal for replaying requests after a crash. Stored Responses are scoped to the API key that created them. - Desktop now loads Qwen3.8-27B correctly from the model library, and TensorFold Web accepts OpenAI-compatible base URLs ending in /v1. Massive thanks to everyone writing code, testing it on their hardware and filing bugs. @AIFlow_ML @andersjw_ @arminxnova @atasrb @Bizuayeu @codengod @devopsfortunato @Eleqtrizit @unigilby @GumbiiDigital @jayleaton @teleogen @lasse_ai @mvanhorn @niklaslenz_ai @petruspennanen, and on GitHub AbrahamCete, BadAd84, jschmied, mrv777, supernovae, ultradaoto, vdalesandro, terrapinelf and Mirrdhyn. I really appreciate the time you all put into this. @petruspennanen @plotarmordev @volatilemarkts @MiaAI_lab @jayleaton
31
26
7
226
12,611
Shoestring Lab retweeted
🚨🚨🚨🚨🚨🚨🚨 USDA HIT. CRITICAL SYSTEMS OFFLINE.
#BREAKING: USDA hit by cyber incident, knocking multiple critical systems offline.
34
168
10
1,152
23,436
MIT researchers found a CPU side-channel called Interrupt Injection that cuts straight through Spectre v2 defenses on both AMD and Intel chips. By arranging a hardware interrupt to arrive precisely at the right moment, unprivileged user-space code can re-poison the branch predictor after the defense has executed but before the CPU relies on the prediction. The timing window exists in the gap between OS implementation and CPU protection. You can pull Linux password hashes from kernel memory despite modern mitigations.
12
45
4
457
19,956
Shoestring Lab retweeted
This one caught only by Kimi-K3.1-Pro! - Opus 5.5 missed it. - Gemini 4 Neon missed it. - GLM 5.3 missed it. - GPT-6.2 Astra missed it. Probably the best open model out there.
37
13
3
525
35,706
Shoestring Lab retweeted
your annual reminder that the LA river is concrete so we don’t all die.
“why is the LA river concrete” because prior to its channelization by the u.s. army corps of engineers the city had regular devastating floods and after one killed 87 people they tried to lynch the mayor.
132
1,358
70
21,762
1,149,353
Shoestring Lab retweeted
Finally some good news !! About Qwen4 ofcourse ! Qwen dev's have been improving Qwen4 so much, thanks to RL that the version we are going to get is a vastly improved version, basically a version bumb or aka Qwen4.1. This is really good, they know they can't release an inferior model, pressure is huge. Now my hopes for Qwen4-27B is even higher !! LFG !!!! Only question is when, but they can take a bit more time if it means we get something that meets our expectations !
btw, qwen4 is a whole different model now from when i first used it like 2 months ago now. it's insane its been this long of a release timeline. it's basically qwen4.1 now with all the extra RL they've been doing
25
19
4
503
30,811
Bannon knew. End H1B.
2
21
1
94
1,296
Shoestring Lab retweeted
Covid money printing caused massive inflation. However, major inflation events like this historically have a tendency to occur in 2 separate waves. First, there's a massive external shock which causes inflation (Covid). Second, years later, everyone realizes their psychological anchors are wrong and the cost of everything has increased. When people catch up psychologically, then a second wave of inflation occurs because roughly simultaneously everyone realizes that their business models are broken because input costs have doubled and now their profit margins are completely inadequate for businesses, and wages are completely inadequate for workers. So then a prolonged second wave of inflation occurs when nobody can afford to pay more but every input cost and labor cost reacts to the original external shock (Covid lockdowns). This is part of a general prolonged stagnation in modern America of a dead zombie economy.
13
27
2
269
6,420
Shoestring Lab retweeted
10
70
1
508
14,658
Shoestring Lab retweeted
Politico reports that White House Chief of Staff Stephen Miller sidelined pro-amnesty Markwayne Mullin from appearing on television after the massive incompetence he displayed during the past week. That’s a great start, but he needs to be fired. We need someone effective at DHS to maintain credibility with the American people and carry out mass deportations.
Markwayne Mullin needs to be fired. He’s so disconnected, or so dumb, that he thought it was OK to fly to a football game while millions of Americans grappled with a national crisis he’s supposed to be helping manage. He’s also totally unfit to be DHS chief because he’s so soft on immigration. He wants more worker visas to “plug holes” in the economy, and he actually told Haitian migrants to fill out the paperwork for permanent status so they could stay here. He has to go.
204
529
40
2,509
108,748
Shoestring Lab retweeted
Maps you can build on.
1
2
76
Shoestring Lab retweeted
Today, we are publishing the 1.5.5 and 1.4.7 security patch versions. Read the blog to learn what changed in each version: cs.co/6011BynU31
5
38
3
315
7,796
Shoestring Lab retweeted
California, I-80 update/ Donner pass- Getting these trucks out is like pulling a frozen woolly mammoth out of the tundra.
41
180
28
789
47,519