@VectorCrossProd

Lead on PCB design software TraceMaker (https://xitter.cf/t.co/QXXc9mLbl9), and LLM visualisation llm-visuals (https://xitter.cf/t.co/nhYsCUxUdh).

 Australia
kookabus.com
Joined July 2026
  • Tweets2.2K
  • Following2.5K
  • Followers1.9K
297Media
Include

Only show posts containing:

Exclude

Hide posts containing:

Time range
-
Minimum likes
Replying to @1k0z4h3dr4
It should handle Ollama models just fine. If it doesn't let me know the issue and I can fix it.
14
Replying to @BHIIKTOR
pedal to the metal!
1
9
Replying to @Compilethings
And the fact it is a preview for Qwen4 is such a good omen!
1
1
86
~100 days ago Grok 4.5 came out and I thought it was a massive breakthrough. Today I'm running Qwen3.8-Flash-Next at Q6 on 10 year old hardware at ~40 tok/s, and on the Artificial Analysis Intelligence Index it scores 40 vs Grok 4.5 (high) at 39. Same chart attached. This is moving so fast.
3
1
262
Replying to @AlexanderJoh10
Fair enough. I want Mistral to succeed. I remember trying in out and liking it waaaaay back in the day but the pace is fast.
1
9
Replying to @loktar00
Yes it is so cool to have ideas come to life.
1
50
Nigel Hungerford-Symes retweeted
沐阳
@yyyoleOct 24
绝了！NVIDIA最新开源了2,575 个高规格 3D 资产，整整472GB！！ 从日常用品到工业设备，里面收录了不少现实环境里的物体。 一揽子，全部放进了 Hugging Face。 NVIDIA SimReady Assets。 绝不单纯是一个“免费建模素材库”！！ NVIDIA给每个3D 资产都带上了物理属性和语义信息，比如有多重，什么材质，重心在哪，用在什么场景等等，全都写进了参数里。 明摆着，就是用来让AI学习的！ 学习怎么和真实的物理世界互动！ 传统的3D资产在AI眼中，就是个有贴图的“视觉资产”，NVIDIA想做的是，把“视觉资产”变成AI可以理解的“物理资产”！ 所以，这批资产的价值可想而知！ 至少能够用到这些场景上！ 机器人抓取和搬运训练，先在虚拟环境里练习拿取、摆放物体； 搭建工厂、仓库的数字孪生，测试设备和物体的布局； 配合合成数据工具，改变光照、视角、遮挡和物体位置，生成带标注的训练数据。 大家都在卷文字、代码与图像生成时，NVIDIA 开始为Physical AI（物理 AI） 铺设地基了。 不过，对大部分人来说，也是一个相当全面的3D资产素材库！ 链接在下面，必须收藏了！
46
402
31
3,688
196,145
Replying to @10_X_eng @dhh
Nice one. looks promising!
1
1
5
Replying to @0xWhiteMage
Just a bit too slow on my hardware for day to day. Great for batch
1
2
478
Replying to @deeperflows
Gotta fill those gaps for humanity! :)
1
1
52
Replying to @Compilethings
Is it just the copying or dies it need pasting too?
1
2
47
Replying to @LouisvilleGun
Another good idea is to turn on PMF on your wifi to avoid deauth attacks
2
184
Replying to @quimedesu
Make stuff up. Insta-mute.
1
3
236
Replying to @richgel999
Same set up. It is good
18
Replying to @EduardoAguirreD @eduardotech
Thanks. Opus 5.5 is still the best I've had for boards, but that's cloud. Locally I'm testing Qwen3.8-Flash-Next Q6 on the R730 now. Which models are you running on the Mac?
32
Replying to @i2cjak
Neither, you are a bot that called a clanker :)
1
336
Replying to @10_X_eng @dhh
Best of luck
1
15
Replying to @will_whang
I would love to see how far TraceMaker could auto route this board: github.com/DingoOz/TraceMake…
1
101
Replying to @10_X_eng @dhh
Do you have the full GPU acceleration working? that is always the big issue with Jetsons and not using Jetpack.
1
2
73
Replying to @alexokita
Thanks, I love to keep it simple with SSH and then max out what is happening. If I can't see my model doing a 5 minute preload, I'll think it is stalled and kill it. So this is really important information with local LLMs
13
Load more