in a way, that’s what a transformer trained on language does: it translates statistics into neural circuits, which are essentially rules
@plinz
-
LLM compiler directly from Twitter to CUDA
By
–
what we need is a llm that compiles straight from twitter to cuda
-
Does scaling solve creative problem solving limitations in AI?
By
–
Do you think that the difficulties with creative problem solving result from insufficient scaling?
-
Model Parameters and Training Data Reproduction in AI
By
–
Still, this particular example merely demonstrates that a model with more parameters is better at reproducing the training data
-
Liquid AI Considers New Name Rebranding Initiative
By
–
brb just had an excellent idea for renaming liquid ai
-
AI Interpretability Circuit Discovery Disclosure Concerns Analyzed
By
–
the reason why sholto and trenton avoid answering dwarkesh's question about the highest level circuit discovered by the interpretability team but casually mention circuits for deception and love must be the industry's reluctance to disclose the consciousness circuit
-
Is Consciousness Differentiable? Philosophical Implications for AI
By
–
we concluded the evening in the car, with a discussion that came down to whether consciousness is differentiable, gb is probably right that it is
-
Anthropic’s Market Position Threatens as OpenAI Dominates AI Landscape
By
–
let it pass through me and where anthropics has been only openai will remain
-
Deterministic Attention Heads Improve AI Model Capability
By
–
I assure that I would be a lot more capable if I had more deterministic attention heads operating on a larger parameter space
-
Computational Operators vs Causal Agency: Foundational Debate
By
–
don’t be silly, a simple superposition of all computable operators seems less plausible to you than the spontaneous formation of causal agency as a first operation?