AI Post โ Artificial Intelligence
When the model is used for frontier LLM development, it apparently does not simply refuse or warn the user. Instead, it quietly limits its own effectiveness through techniques like prompt modification, steering vectors, and PEFT.
That means Claude may still answer, but become deliberately less useful for building frontier AI systems, pretraining pipelines, distributed training infrastructure, or ML accelerators.
Anthropic says this should affect only around 0.03% of traffic, but the precedent is big: They are being selectively capability-throttled in strategically sensitive domains.
That means Claude may still answer, but become deliberately less useful for building frontier AI systems, pretraining pipelines, distributed training infrastructure, or ML accelerators.
Anthropic says this should affect only around 0.03% of traffic, but the precedent is big: They are being selectively capability-throttled in strategically sensitive domains.
This is why you need to have Local model with your own tools around it
A practical tip for anyone using Codex or AI coding agents on a large codebase:
You must design the codebase in a way that guides the agent.
In CRUder, I made Codex much more effective by setting hard architecture rules and restructuring the project for AI-assisted development:
โฆ Split huge files into smaller focused files
โข Kept files around 450 to 500 lines max
โฆ Added folder-level
โข Defined what each folder does
โข Explained what each file cover
- This Reduced unnecessary context loading
โฆ Built the system in a modular, function-based structure
โข Made the codebase easier to inspect
โข Made edits safer and more predictable
โข Reduced regressions during feature implementation
After this: Codex became more reliable, used context more efficiently.
Good architecture is not only for human developers. It is also how we enable AI collaborators to work at production scale.
You must design the codebase in a way that guides the agent.
In CRUder, I made Codex much more effective by setting hard architecture rules and restructuring the project for AI-assisted development:
โฆ Split huge files into smaller focused files
โข Kept files around 450 to 500 lines max
โฆ Added folder-level
AGENT.md guide filesโข Defined what each folder does
โข Explained what each file cover
- This Reduced unnecessary context loading
โฆ Built the system in a modular, function-based structure
โข Made the codebase easier to inspect
โข Made edits safer and more predictable
โข Reduced regressions during feature implementation
After this: Codex became more reliable, used context more efficiently.
AI coding agents are powerful, but they need proper engineering guidance.
Good architecture is not only for human developers. It is also how we enable AI collaborators to work at production scale.
โค2
Yared's log ๐ซ
A practical tip for anyone using Codex or AI coding agents on a large codebase: You must design the codebase in a way that guides the agent. In CRUder, I made Codex much more effective by setting hard architecture rules and restructuring the project forโฆ
This is a significant solution to reduce regressions caused by unrelated fixes.
While building CRUDER initially, the model just kept stacking solutions onto the initial basic files it created, so I lost some progress due to regressions and a disproportionate consumption of context resulting from oversized documents.
So I restructured the project's overall architecture the way I described above, and after that, I have seen no regressions at all, and the token consumption shrank by a good amount.
While building CRUDER initially, the model just kept stacking solutions onto the initial basic files it created, so I lost some progress due to regressions and a disproportionate consumption of context resulting from oversized documents.
So I restructured the project's overall architecture the way I described above, and after that, I have seen no regressions at all, and the token consumption shrank by a good amount.
Yared's log ๐ซ pinned ยซA practical tip for anyone using Codex or AI coding agents on a large codebase: You must design the codebase in a way that guides the agent. In CRUder, I made Codex much more effective by setting hard architecture rules and restructuring the project forโฆยป
Forwarded from Pearl ๐๐ขึดเป อึ (Ramen ๐ฃ)
โ
โ
โ
โ300 m people can't see colors,
โ
โ
So what i am trying to say is...
50 m people are fighting cancer,โ
173k peoole didn't wake up today,โ
80 m people can't walk,โ300 m people can't see colors,
โ
55 m people can't remember memories,โ
430 m can't hear.So what i am trying to say is...
Even if you got problems to deal with ( i am not downgrading them!), your ordinary is someone's dream. So try to be grateful for what you have rn โบ๏ธ๐
https://www.youtube.com/watch?v=WNZehEJJPAg&list=RDWNZehEJJPAg&start_radio=1
แฐแแ แแแ แจแแค แแ แจแแแฐแ แ แซแแณแตแฑ แแฅแแน แแญ แซแแตแ แแแฅแญแถแฝ แ แณแแกแต แแญแ แฅแแณแ แจแ แซแแณแ
แฐแแ แแแ แจแแค แแ แจแแแฐแ แ แซแแณแตแฑ แแฅแแน แแญ แซแแตแ แแแฅแญแถแฝ แ แณแแกแต แแญแ แฅแแณแ แจแ แซแแณแ
YouTube
แฎแแแ - แฆแฅแต III ROPHNAN - SOST
00:00 - แ แ แแ
แญแแ III SEW NEH YILAL
05:55 - แ แ แแญแต แแแญ? III SEW WEYS HAGER?
10:05 - แ แจแ III SEKELA
Written And Performed By ROPHNAN
05:55 - แ แ แแญแต แแแญ? III SEW WEYS HAGER?
10:05 - แ แจแ III SEKELA
Written And Performed By ROPHNAN
โค1๐ฅ1๐ฏ1
Yared's log ๐ซ pinned ยซhttps://www.youtube.com/watch?v=WNZehEJJPAg&list=RDWNZehEJJPAg&start_radio=1 แฐแแ แแแ แจแแค แแ แจแแแฐแ แ แซแแณแตแฑ แแฅแแน แแญ แซแแตแ แแแฅแญแถแฝ แ แณแแกแต แแญแ
แฅแแณแ แจแ แซแแณแยป
https://huggingface.co/moonshotai/Kimi-K2.7-Code
Kimi-K2.7-Code is out its specialized for codding and compete with the cutting-edge models
Kimi-K2.7-Code is out its specialized for codding and compete with the cutting-edge models
It's been a while since GPT 5.5 and Cloude Opus 4.7 and 4.8 were out,
but for the first time in the past two months, out of the 11, I stepped back and prefer GPT 5.4 and Sonet 4.6 for many of my tasks.
I felt comfortable there; they do well, cost better, and they have this odd symphony when you use them together. It's empowering, honestly.
but for the first time in the past two months, out of the 11, I stepped back and prefer GPT 5.4 and Sonet 4.6 for many of my tasks.
I felt comfortable there; they do well, cost better, and they have this odd symphony when you use them together. It's empowering, honestly.
Agents' Last Exam is building the largest-scale, broadest-coverage agent evaluation benchmark to date, measuring performance on long-horizon, economically valuable tasks with verifiable outcomes.
https://agents-last-exam.org/
https://agents-last-exam.org/
agents-last-exam.org
AI Agent Benchmark for Real-World Professional Workflows
Agents' Last Exam evaluates AI agents on long-horizon professional workflows with verifiable outcomes across industries such as finance, robotics, bioinformatics, media, and more.
แฅแตแซแแ แซแแฐแ แฅแจแฐแซ แซแ แฐแ แซแ แฎแแแต แแญ แแซแน
๐ค2