Also, at the point nukes were regulated, their power was not nearly entirely in the hands of profit-seeking entities which would benefit financially from that regulation — unlike the case with AI today.
@jeremyphoward
-
Fast.ai Model Achieved 100% Accuracy in Classification Task
By
–
We used this example in http://
fast.ai class years ago and even then it got 100% correct! -
FSDP vs DDP Communication Overhead Derivation Explained
By
–
Anyone know how to derive this '1.5x' communication overhead between FSDP vs DDP (from the FSDP paper)?
-
Continuous batching approach and batch size impact discussion
By
–
Yes, and their continuous batching approach – I figured it's largely orthogonal, although I wondered if it makes less of a difference there since it's not really a batch-size of 1 generally.
-
vLLM Performance Comparison and Optimization Approaches
By
–
Thanks for sharing this. How does performance compare to vllm's approach? Can the two approaches be combined?
-
Creator Monetization Strategy: Blue Check Worth Assessment
By
–
Yeah I signed up so I'd get paid, but in hindsight it wasn't worth the effort. (I turned off the blue check thing because it seems pretty tacky to me.)
-
Example from recent paper not in training data
By
–
Here's an example from a paper that came out last month – so it wasn't trained on it.
-
ChatGPT Completes Tests Remarkably Fast During Writing
By
–
Cool in the time it took to write that ChatGPT finished running its tests.
-
GPT-4 Recognition of Mathematical Equations Cross-Entropy Adam
By
–
I should note that these example equations are pretty famous – they're for cross-entropy and adam optimiser; obviously GPT4 recognised them, since it named the functions correctly, although I never said what they were. Perhaps for less-known equations it would be less good.
