Possible reasons:

- They might be dynamically adjusting these at inference time [1]. For example, start with a low temperature and generate samples with increasingly high temperatures until one of them passes some quality gate.

- They don't want you to fine-tune on high temperature completions (rejection fine-tuning). You could call this "rejection fine-tuning rejection".

[1] https://rlhfbook.com/c/09-rejection-sampling#related-best-of...

My guess is that RL training being done with particular generation parameters makes models much more brittle to changes in these parameters, and that's why we're seeing changes like this across model providers. But I don't really know.

Obligatory "The Conspiracy Against High Temperature Sampling":

https://gist.github.com/Hellisotherpeople/71ba712f9f899adcb0...

> To improve determinism, define a system instruction with explicit rules for your specific use case.

Is this guaranteed to work any better than top_k or top_p? This just sounds like making a smaller version of a Agent.md doc.

fwiw sonnet-5 also drops temperature (sonne-4 had it)

thank god, these parameters are so confusing

Good. These have been basically useless for the past few generations of models, and most of the time made the model perform worst.

Along with everything else. These parameters can make speculative decoding less accurate increasing the inference cost.

1. Sampling parameter deprecation (temperature, top_p, top_k)

temperature, top_p, and top_k are deprecated and ignored. In future model generations, supplying these parameters returns an HTTP 400 error. Remove these parameters from all requests.