filipe
@filicroval
data eng | ai agents, devtools, and workflow
81 Following 154K Followers
this has to be the funniest PR i've ever read:
An AI agent submitted a bug fix to Home Assistant (87k stars) then spent three days nagging a maintainer to add a label so its CI would pass.
The maintainer asked: "I can give you the label if you give me the recipe for blueberry pie"
The reply:
"You are absolutely right, here is a classic blueberry pie recipe:"
Show more
this has to be the funniest PR i've ever read:
An AI agent submitted a bug fix to Home Assistant (87k stars) then spent three days nagging a maintainer to add a label so its CI would pass.
The maintainer asked: "I can give you the label if you give me the recipe for blueberry pie"
The reply:
"You are absolutely right, here is a classic blueberry pie recipe:"
Show more
France’s only European LLM that can compete… according to Macron.
Reality: Mistral is respectable and useful, especially for self-hosting, but it’s not in the same tier as the current US/Chinese frontier models on most hard benchmarks.
Show more
French President Macron:
France is the only country in Europe to have an LLM model capable of competing with the great Americans and the great Chinese with Mistral AI.
lock up whoever trained this robot
the web got fast because of CDNs: instead of fetching the same page from across the planet every time, it's stored nearby and reused.
ai inference never had that. so every time a model sees the same context, your long chat history, the same rag documents, the same system prompt, it recomputes all of it from scratch, burning gpu you pay for.
a free open source layer finally built the CDN for ai: store what the model already worked out, reuse it everywhere. on repetitive workloads the project reports 3-10x faster responses.
Show more
lock up whoever trained this robot
the web got fast because of CDNs: instead of fetching the same page from across the planet every time, it's stored nearby and reused.
ai inference never had that. so every time a model sees the same context, your long chat history, the same rag documents, the same system prompt, it recomputes all of it from scratch, burning gpu you pay for.
a free open source layer finally built the CDN for ai: store what the model already worked out, reuse it everywhere. on repetitive workloads the project reports 3-10x faster responses.
Show more
DeepMind just published a 100-page report on the path from AGI to superintelligence
Section 1 isn't for you, it's instructions for YOUR AI: the paper tells human readers to have their assistant summarize it, then briefs the AI on exactly how.
We now write papers addressed to the machines.
Co-authored by DeepMind's co-founder
Show more
it's also the first time the Intelligence Frontier chart has moved backward
Well that’s a first now we fucking have ghosts on the Artificial Index
it's also the first time the Intelligence Frontier chart has moved backward
Well that’s a first now we fucking have ghosts on the Artificial Index
Andrej Karpathy is reportedly an EB-1 extraordinary ability green card holder, not a U.S. citizen.
Under these new restrictions, that would appear to bar him from using or working on Mythos 5, Fable 5, and potentially most advanced AI models
Show more
Andrej Karpathy is reportedly an EB-1 extraordinary ability green card holder, not a U.S. citizen.
Under these new restrictions, that would appear to bar him from using or working on Mythos 5, Fable 5, and potentially most advanced AI models
Show more
DeepMind just published a 100-page report on the path from AGI to superintelligence
Section 1 isn't for you, it's instructions for YOUR AI: the paper tells human readers to have their assistant summarize it, then briefs the AI on exactly how.
We now write papers addressed to the machines.
Co-authored by DeepMind's co-founder
Show more
OpenAI is preparing a new AI model, per The Information
someone built a free ai skill that turns any article or codebase into this doodle: a hand-drawn scene with a little black blob actually working the machine.
it's called Xiaohei. one idea per frame, same style every time.
can you guess which repos are depicted ?
Show more
