Tensor Templar @TensorTemplar

Chief Intellectrician (MLRE & EE), decoupling productivity from human labor with AI. Nuke and Compute maximalist. Tensortemplar.substack.com Joined February 2022

Tweets

7K
Followers

249
Following

709
Likes

21K

Tensor Templar @TensorTemplar

4 days ago

Guess which ~4 trillion USD market cap company has quality charts like these on the official product page. If you are a chart sloponaught with wild theories - i would love to hear them.

0 0 2 16K 0

Download Image

"Once open models exceed these arbitrary thresholds — as they have in the EU — developers could face uncertain and indeterminate liability, with no clear path to compliance." This is exactly the kind of thing that I wanted to avoid when moving my family to a foreign country.

AI Frontiers @aif_media

4 days ago

8 29 126 139K 53

0 0 1 63 0

Tensor Templar @TensorTemplar

5 days ago

Wake up goto kitchen Wife greets with: look who is so attractive today! Static-charged grinder

0 0 1 30 0

Andrej Karpathy @karpathy

7 months ago

For friends of open source: imo the highest leverage thing you can do is help construct a high diversity of RL environments that help elicit LLM cognitive strategies. To build a gym of sorts. This is a highly parallelizable task, which favors a large community of collaborators.

322 846 9K 1.2M 3K

Ebba Busch @BuschEbba

2 weeks ago

Sweden joins the global nuclear race! Today marks the start for Sweden in the global nuclear power race. Vattenfall's decision to take the next step to build new nuclear power is not just a choice of path for our country's future - it is the start of something bigger. Sweden…

171 269 2K 173K 69

Download Image

DeepSeek @deepseek_ai

2 weeks ago

Model Update 🤖 🔹 V3.1 Base: 840B tokens continued pretraining for long context extension on top of V3 🔹 Tokenizer & chat template updated — new tokenizer config: huggingface.co/deepseek-ai/De… 🔗 V3.1 Base Open-source weights: huggingface.co/deepseek-ai/De… 🔗 V3.1 Open-source weights:…

16 45 837 138K 64

ARC Prize @arcprize

3 weeks ago

Analyzing the Hierarchical Reasoning Model by @makingAGI We verified scores on hidden tasks, ran ablations, and found that performance comes from an unexpected source ARC-AGI Semi Private Scores: * ARC-AGI-1: 32% * ARC-AGI-2: 2% Our 4 findings:

39 203 1K 266K 600

Download Image

Tensor Templar @TensorTemplar

a month ago

Weekend ruined, plans cancelled. What is an efficient way to update router metrics without calculating them each step?

1 0 11 32K 0

Download Image

Tensor Templar @TensorTemplar

2 months ago

If your weekend is boring so far, here is what to do

ARC Prize @arcprize

2 months ago

If your weekend is boring so far, here is what to do

71 228 2K 368K 536

Download Image

0 0 0 90 0

Nathan Lambert @natolambert

2 months ago

It is a major policy failure that the US cannot accommodate top AI conferences due to visa issues.

48 163 1K 326K 114

Download Image

Tensor Templar @TensorTemplar

2 months ago

This thread makes me update my timelines on xAI's progress more than the public benchmarks. Meanwhile everyone else is worried about sycophancy

Mr Hills (Пагорбні простріли) @throught_hills

2 months ago

This thread makes me update my timelines on xAI's progress more than the public benchmarks. Meanwhile everyone else is worried about sycophancy

133 1K 6K 710K 1K

Download Image

0 0 1 78 0

Tensor Templar @TensorTemplar

2 months ago

We need coal irons, but for swapping residential main fuses for european VRE grids. Who is working on this?

Tensor Templar @TensorTemplar

2 months ago

We need coal irons, but for swapping residential main fuses for european VRE grids. Who is working on this?

0 0 1 95 0

Download Image

0 0 0 56 0

Tensor Templar @TensorTemplar

2 months ago

Just a normal day in the life between inertia and a grid-scale PV when the clouds are moving fast. Totally normal voltage spikes that on my low voltage lines. Would suck to have sensitive equipment grid parallel

0 0 1 95 0

Download Image

@levelsio @levelsio

2 months ago

No, that was because Russia funded extreme left activists in Germany to shutdown nuclear plants so that they would become fully dependent on Russian gas Not some theory either, it was published by the European Parliament I seriously bet the anti-AC eco mindvirus similarly is…

Alex 🔔 @_avdept

2 months ago

5 2 10 231K 1

119 208 2K 232K 569

Download Image

Tensor Templar @TensorTemplar

2 months ago

Gemini's circuits returned frustration and it chose to experience it. Instantly bonding over HF dependency management hell.

0 0 0 40 0

Download Image

Tensor Templar @TensorTemplar

2 months ago

I really hate what LLMs did to the field and I am not talking about LLM generated papers. I am sure to not be alone in this, so here are some simple guidelines to decide if you should submit your paper and how to improve it: 1. If you are using private data, private training…

0 0 7 13K 1

Tensor Templar @TensorTemplar

3 months ago

Had a similar experiment going but they beat me to it. Please print it out and slam on decel desks ins your area to restore sanity to researcher TLs

Rohan Paul @rohanpaul_ai

3 months ago

Had a similar experiment going but they beat me to it. Please print it out and slam on decel desks ins your area to restore sanity to researcher TLs

97 324 2K 475K 1K

Download Image

0 0 0 95 0

Tensor Templar @TensorTemplar

3 months ago

You can differentiate 90% of the ML yappers on Twitter by showing them this plot of a MoE pre-training run and asking for how many of their monthly salaries spent each day this job should be kept running with the lab's runway chart on the screen next to it. If they don't flinch…

vik @vikhyatk

3 months ago

27 7 334 44K 127

0 0 0 74 0

Download Image

ispace @ispace_inc

3 months ago

Fly me to the Moon 🎵🌝 RESILIENCE status: nominal Distance above the Lunar surface: ca. 100 km Current orbital phase: Low lunar orbit, traveling at ca. 5,800 km/h RESILIENCE remains in a circular orbit as landing day approaches. This video was captured from lunar orbit by…

37 382 2K 132K 113

Download Video

Anthropic @AnthropicAI

3 months ago

Our interpretability team recently released research that traced the thoughts of a large language model. Now we’re open-sourcing the method. Researchers can generate “attribution graphs” like those in our study, and explore them interactively.