True – I meant it doesn't work ootb without code changes. Hopefully popular models will be updated in the future to work correctly with the meta device.
@jeremyphoward
-
LLM Buffer Limitations: Non-Persistent Memory Architecture Challenges
By
–
That doesn't work for anything with non-persistent buffers (which is most current LLMs AFAICT, since the sin/cos cache is in those buffers)
-
Unfettered Utilitarianism: A Recipe for AI Evil
By
–
Because unfettered utilitarianism is a recipe for evil
-
EA’s Gap Between Aspiration and Reality Explained
By
–
What EA aspires to be, and what it’s become, are very different things. This difference was entirely predictable.
-
RAG Accuracy Issues in AI Systems Explained
By
–
Hey don't put me in the middle – lowest RAG accuracy there! :O
-
Math Updates Required for Next Generation AI
By
–
Seems the math needs to be updated for nextgen https://
x.com/choldgraf/stat
us/1727819994640289935
… -
Ethics Beyond Utilitarianism: Balancing AI Decision-Making
By
–
For sure but they were never unfettered utilitarianists. EA has good aspects but you gotta actually use some common sense and listen to your inner voice too.
-
Philosophy of AI: Examining 30-Year-Old Undergraduate Concepts
By
–
I remember learning this 30 years ago in undergraduate philosophy. And it was not at all novel at that time.
-
Unfettered Utilitarianism in AI Creates Negative Consequences
By
–
Yeah a lot of people unreasonably surprised that unfettered utilitarianism turns everything to shit
-
A100 vs 4090: GPU Availability and Cost Comparison
By
–
However many a100s you can buy, you can buy far more 4090s