Greg Slepak (@[email protected])

44.1K posts

Greg Slepak (@taoeffect@crib.social) banner
Greg Slepak (@taoeffect@crib.social)

Greg Slepak (@[email protected])

@taoeffect

@okTurtles @Group_Income @espionageapp #Fediverse: https://t.co/hYJJrmMskX

Katılım Eylül 2008
3K Takip Edilen3.4K Takipçiler
zR
zR@zRdianjiao·
🚀 GLM-5.1-HighSpeed is live: 400 tokens/s — a new speed ceiling for flagship-tier LLM APIs. Not a smaller model traded for speed. A flagship from @Zai_org that's also the fastest. 📖 Full technical deep-dive 👇 tilert.ai/blog/speed-as-…
GIF
English
44
77
951
66.7K
Greg Slepak (@taoeffect@crib.social)
Being very good at your job will teach you an important meta-skill called observation. Whether you're a programmer, or a scientist, or a lawyer, a doctor, a police officer, or a chef, to be great at your job you need to understand *DETAILS*. "The devil's in the DETAILS."
English
0
0
0
42
Dylan Avery
Dylan Avery@iamdylanavery·
he's literally doing the Miles Dyson speech from Terminator 2
English
1
0
0
61
Tongyi Lab
Tongyi Lab@Ali_TongyiLab·
Meet Qwen3.7-Max, our latest proprietary model built specifically for the agent era. • Frontier Coding Agent: Excels from frontend prototyping to complex software engineering. • Cowork Productivity Assistant: Reliable office assistant and workflow automation via MCP and multi-agent orchestration. • Long-Horizon Autonomous Execution: Built for sustained, autonomous execution across long-horizon tasks. • Cross-Harness Generalization: Consistent performance across diverse agent frameworks. We designed it to be a versatile agent foundation—equally capable of writing/debugging code, automating office workflows, and sustaining autonomous execution across hundreds or thousands of steps. 👇
English
25
68
597
7.1M
Greg Slepak (@taoeffect@crib.social)
Dez 0.5.0 released! • Adds support for multiple cursors • Codebase refactored to SBP + MVC • Hopefully fixes a bug where [AGENT] pills go changed to [USER] pills
English
1
0
0
59
Greg Slepak (@taoeffect@crib.social)
How do you know when you're ready to actually use AI? Simple: pick the best AI and have it do some of your work. If you think what it does is brilliant, you're not ready. If you think it's retarded, you're ready.
English
0
0
0
34
Greg Slepak (@taoeffect@crib.social)
However, once you have finished your learnings, once you are *better* than the best of the AIs out there, then the AIs can be a very useful tool to save you time. Before then, if you want to write good software, avoid AI. Learn languages. Build your experience.
English
1
0
0
55
Greg Slepak (@taoeffect@crib.social)
Some fledgling software developers are — rightly — concerned about AI and do not like it. Indeed, they should dislike it, because it is an enemy to their own learning of software development.
English
1
0
0
90
Greg Slepak (@taoeffect@crib.social)
You’re adjusting your attention heads to focus on the Ground Truth, using Grace as a regularization technique to prevent overfitting to worldly noise, and treating repentance as a daily process of zeroing out your gradients.
English
1
0
0
43
Greg Slepak (@taoeffect@crib.social)
Accepting Jesus Christ as your personal Lord and Savior is about aligning your model with God. It's about fine-tuning your human. It’s basically RLHF, but the “H” stands for Heavenly.
English
1
0
0
53
Greg Slepak (@taoeffect@crib.social)
Don't talk to me about LLM hallucinations when they're being trained to "align" with human hallucinations.
English
1
0
0
51
Greg Slepak (@taoeffect@crib.social)
@cormachayden_ It's not just the food in stores that contaminated, it's your "healthy" orders of "organic ____ powder" and other health foods that are contaminated too. I tested some from @Amazon and it contained high levels of lead, coliforms, aerobic bacteria, yeast & mold, and chlorpyrifos.
English
0
0
0
38
Cormac
Cormac@cormachayden_·
big water is trying to silence us
English
115
887
3.1K
122.2K