Former researcher says Anthropic can retrain Mythos with far less compute
Original title: Anthropic can already retrain Mythos with "a fraction of the compute originally needed," says former researcher Coxon
Jacob Coxon says an IPO roadshow slide on accelerating efficiency led him to quit after about four months.
Former OpenAI and Anthropic researcher Jacob Coxon says a compute chart in Anthropic’s IPO roadshow was a major reason he left after about four months. He argues Mythos, a spring capability jump, sits on a steepening efficiency curve and could already be retrained with a fraction of the original compute. Anthropic says the points are internal estimates for illustration only. Coxon wants frontier development paused next year.
Unconfirmed: Coxon’s reading of the slide is his own; Anthropic says the figures are illustrative internal estimates.
- The chart’s y-axis is logarithmic and labeled Pre-training compute multiplier.
- Coxon says the climb runs from July 2024 to October 2026.
- He argues algorithmic improvements are on the verge of being automated.
- A pause, he writes, would mainly harm OpenAI and Anthropic.
- AI models
- Mythos
From the original article
2 min read at the-decoder.com
Former OpenAI and Anthropic researcher Jacob Coxon points to a slide from Anthropic's IPO roadshow as a reason he left, saying it shows the company's algorithms are improving not just quickly but at an accelerating rate. Coxon continues to call for a pause on frontier model development so competitors can catch up and human-level AI can become a commodity.
By Matthias Bastian · the-decoder.com
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.