« LongVILA » : différence entre les versions
(Page créée avec « ==en construction== == Définition == XXXXXXXXX == Français == ''' LongVILA''' == Anglais == '''LongVILA''' A comprehensive framework that enables vision-language models to perform complex reasoning on long videos using reinforcement learning. The work addresses the significant challenge of understanding hour-long videos that require temporal, spatial, goal-oriented, and narrative reasoning capabilities. A framework for scaling vision-language models... ») |
Aucun résumé des modifications |
||
| Ligne 20 : | Ligne 20 : | ||
[[Catégorie: | [[Catégorie:vocabulary] | ||
Dernière version du 20 septembre 2025 à 10:35
en construction
Définition
XXXXXXXXX
Français
LongVILA
Anglais
LongVILA
A comprehensive framework that enables vision-language models to perform complex reasoning on long videos using reinforcement learning. The work addresses the significant challenge of understanding hour-long videos that require temporal, spatial, goal-oriented, and narrative reasoning capabilities. A framework for scaling vision-language models to long videos using reinforcement learning, achieving strong performance on various reasoning tasks with a specialized training infrastructure.
Source
[[Catégorie:vocabulary]
Contributeurs: wiki
