It’s not just the narrow wavelength. Even with a perfectly monochromatic green light, your green receptors would activate a lot but your receptors for red and blue would still activate a bit. These researchers specifically target only the green receptors to activate (by literally shooting light at those receptors in particular), so for the first time ever your brain reads a pure green signal.

tariff = (export - import) / (<??? factor> × <another ??? factor that cancels the first one out> × import)

The only times anyone would use the asterisk as multiplication symbol are

• they are doing some fancy math and it’s not the same kind of number multiplication we’re familiar with
• they are on a computer, the keyboard does not have a (×) key, and they don’t know how to typeset it (\times in LaTex), so they just use the asterisk instead

The US government falls in the second category.

It kinda works you just gotta be careful with what you use and keep some human in the loop curating the outputs.

This meme was about training on model outputs. But would be nice if they got some trade secrets as well. Intellectual property is cancer and these IP-stealing Chinese companies, if they exist, are doing god’s work 😊 hope Indian companies steal from China next as well

The ball can quantum mechanically tunnel out to the true minimum. In this sense the local minimum is actually not perfectly stable.

Okay that sounds like the best one could get without self-hosting. Shame they don’t have the latest open-weight models, but I’ll try it out nonetheless.

Interesting. So they mix the requests between all DDG users before sending them to “underlying model providers”. The providers like OAI and Anthropic will likely log the requests, but mixing is still a big step forward. My question is what do they do with the open-weight models? Do they also use some external inference provider that may log the requests? Or does DDG control the inference process?

Stop depending on these proprietary LLMs. Go to !localllama@sh.itjust.works.

There are open-source LLMs you can run on your own computer if you have a powerful GPU. Models like OLMo and Falcon are made by true non-profits and universities, and they reach GPT-3.5 level of capability.

There are also open-weight models that you can run locally and fine-tune to your liking (although these don’t have open-source training data or code). The best of these (Alibaba’s Qwen, Meta’s llama, Mistral, Deepseek, etc.) match and sometimes exceed GPT 4o capabilities.