Mistral-Large-4
Mistral AIWe optimize a highly heterogeneous training mixture covering multimodal, alignment, instruction-following, cybersecurity, code agents, knowledge work, finance, STEM, safety, long-context, and multilingual domains.
general family · filed under data curation
A broad training mixture combining heterogeneous domains or data types, with no more specific shared sampling procedure established.
The sources on this page only mention it in passing. The line above is the extraction's one-line gloss. See data mixture & curriculum for the methods filed beside it.
Also called Heterogeneous training data.
2 quotes from 2 sources, one card per model, each opening with its strongest.
We optimize a highly heterogeneous training mixture covering multimodal, alignment, instruction-following, cybersecurity, code agents, knowledge work, finance, STEM, safety, long-context, and multilingual domains.
we developed techniques to effectively learn from a diverse assortment of real-world robot data, human video data, and synthetic data