Viewing profile — stri8ed

It's not a conspiracy. There's a finite amount of compute available, and they will sell it to the highest bidder. If another company can produce the same intelligence for cheaper, …

Output cost is 3x from Gemini 3 flash.

Given the cost increase associated with this model, and previous model releases, I think the size is trending upwards, not down.

Is it fair to characterize a founder of a failed startup as a con man? I.e. did he make claims that were factually untrue, and intentionally deceived investors?

What is the counterfactual? Without knowing the number of attacks prevented by these tools, we don't know what the baseline would be.

Having adequate law enforcement training and funding, is not mutually exclusive with leveraging technology for more effective enforcement. In fact that's where some of the funding …

[flagged]

Not a chance. Even if American companies did abide by it, there is no reason Chinese companies would. And good luck definitely proving that a model trained on it.

That problem along with its many solutions are surely littered throughout the training data. Not to mention, it would be trivial to overfit on that problem. I don't know why people…

It's a result of the system prompt, not the base model itself. Arguably, this just demonstrates that the model is very steerable, which is a good thing.

Trained at least in part on Chat-GPT data.

It would happen in China regardless what is done here. Removing billionaires does not fix this. The ship has sailed.

The benchmarks agree as well.

Isn't that how previous models were, before the attention is all you need paper?

Or be in the business of building infrastructure for AI inference.

You can modify this, by setting the GPT4 system prompt, with instructions of your preferred response style.

For programming, GPT4+. I was excited to switch to Claude after hearing all the positive anecdotes. Having tried it, I'm very unimpressed. It spouted complete, confident sounding n…

It's impossible to prove.

I don't think you can infer someone's age and skin colon from that post.

In humans, a software problem can become a hardware problem, due to plasticity in the hardware. See alcoholism for example.

Do existing LLM's not already train on this data?

In such a scheme, wouldn't synonyms of the same word be no closer to each other, than any other random string?

What is the rate of increase, and how does it compare to CPI inflation?

How does the quantization happen? Are the weights preprocessed before loading the model?