Keep on to blur preview images; turn off to show them clearly
RT @FaroukAdeleke3: Holistic Benchmarking with dspy-helm from Asad Aali on @modaicdev ! ☺️ Paper: https://t.co/Imo1UXkek7 Github: https:/…
Asst professor @MIT EECS & CSAIL (@nlp_mit). Author of https://t.co/VgyLxl0oa1 and https://t.co/ZZaSzaRaZ7 (@DSPyOSS). Prev: CS PhD @StanfordNLP. Research @Databricks.
What's remarkable is that general intelligence emerges evolutionarily from the simple goal "you need to survive/succeed through a series of ever-novel, often adversarial situations" General intelligence is a situated process of efficient adaptation to novelty
Co-founder @ndea. Co-founder @arcprize. Creator of Keras and ARC-AGI. Author of 'Deep Learning with Python'.
Because our universe follows stable laws, a sufficiently general intelligent system adapted to it, like human-driven science, can eventually model any phenomenon within it. Human intelligence may not be "universal" in the mathematical sense (see No Free Lunch theorem), but we are perfectly adapted to decode the specific language of our universe. That's enough for us to be "generally intelligent".
What's remarkable is that general intelligence emerges evolutionarily from the simple goal "you need to survive/succeed through a series of ever-novel, often adversarial situations" General intelligence is a situated process of efficient adaptation to novelty
V3.2 is the best open model on SWE-Rebench, but by a hair, and not the most cost-efficient… until you realize that they haven't used caching for it. Costs of agentic coding are completely dominated by prefill. With ≈90% cache hits, Whale would fall to ≈$0.1 per problem.
Of course DeepSeek has caching at home. At pass@5 70.2 they are still at Opus 4.5 level at a tiny fraction of the cost, and only really lose to GPT 5.2 and well-designed tool systems. This should give you an idea of the next gen's pass@1 ceiling.
the ai infra buildout is injecting enormous liquidity into american infrastructure. it needs to be allowed to flow into energy production and transmission. the way to do that is to charge datacenters more to be connected faster. it's that simple.
we are going to unlock american terawatts.
RT @thejustinwelsh: 99% of the internet is consuming. 1% of the internet is creating. The 99% pay the 1% with their time, attention, and m…
The $10M Solopreneur | Helping 100,000+ experts turn their expertise into income.