Open Models Almanac
Methodsmodel architecturemultimodal architecture

specific method · filed under model architecture

Mixed-modality training from the start

Mixed-modality training from the start trains a model on text, image, and video together from the very first step rather than adding modalities later.1 Training all modalities jointly from the beginning enables deeper semantic fusion across them.2

How this definition was written

Summarised by glm-5.3-flash from 1 source; every sentence is checked against the quote it cites.

  1. M3 undergoes mixed-modality training from the very first step, enabling deeper semantic fusion across text, image, and video — README
  2. enabling deeper semantic fusion across text, image, and video — README

Also called Mixed-modality training, mixed modalities from the start.

sources
4
model
1
lab adopt it
1
strongest
core

Documented in

MiniMax

Evidence

4 quotes from 4 sources, one card per model, each opening with its strongest.

How sources treat it
core 4

MiniMax-M3

MiniMaxadopted4 sources

M3 is a model that has undergone mixed-modality training from Step 0.

coretraining objectivein MiniMax-M3MiniMax
3 more from this model

MiniMax says M3 was trained with mixed modalities from the start.

coretraining objectivein MiniMax-M3MiniMax

M3 undergoes mixed-modality training from the very first step

coretraining objectivein MiniMax-M3MiniMax

M3 undergoes mixed-modality training from the very first step

coretraining objectivein MiniMax-M3MiniMax

Further reading

Picked by hand, not extracted: where to read more, not evidence for anything on this page.

↑