mistralai

/

Devstral-2-123B-Instruct-2512

like 250

Follow

Mistral AI_ 14.2k

Hello,

This PR adds the required hparam arguments to enable stochastic sampling (temp 0.15) rather than greedy decoding in the generation_config.json.
So when users load the Devstral-2-123B-Instruct-2512 model, they automatically get the default sampling settings intended by Mistral.

Motivation: Not all users might know about these sampling hparams and what they do, defaulting to what Mistral recommends, could lower complaints about potential poor generations/model performances.

Since there was "_from_model_config": true, I checked config.jsonfirst but there were no sampling hparam declared there, hence my changes here.

Opened as a separate PR in case you want to keep it greedy by default, this is linked to: https://huggingface.co/mistralai/Devstral-2-123B-Instruct-2512/discussions/9
(It's true that low temp is close to greedy but this is still stochastic nonetheless)