« KRIS-Bench » : différence entre les versions
(Page créée avec « ==en construction== == Définition == XXXXXXXXX == Français == ''' KRIS-Bench''' == Anglais == '''KRIS-Bench''' A comprehensive benchmark for evaluating knowledge-based reasoning capabilities in image editing models. While recent advances in multi-modal generative models have improved visual quality in image editing, their ability to perform edits that require real-world knowledge and reasoning remains under-explored. KRIS-Bench addresses this gap by provid... ») |
Aucun résumé des modifications |
||
| Ligne 15 : | Ligne 15 : | ||
[https://huggingface.co/papers/2505.16707 Source : huggingface] | [https://huggingface.co/papers/2505.16707 Source : huggingface] | ||
[https://www.chatpaper.ai/fr/dashboard/paper/ca6d507d-3e59-4c3d-91bc-45205ed57bf8 Source : ChatPaper.ai] | |||
[[Catégorie:vocabulary]] | [[Catégorie:vocabulary]] | ||
Version du 11 juin 2025 à 14:52
en construction
Définition
XXXXXXXXX
Français
KRIS-Bench
Anglais
KRIS-Bench
A comprehensive benchmark for evaluating knowledge-based reasoning capabilities in image editing models. While recent advances in multi-modal generative models have improved visual quality in image editing, their ability to perform edits that require real-world knowledge and reasoning remains under-explored. KRIS-Bench addresses this gap by providing a structured framework to assess how well models can apply different types of knowledge during the editing process.
Source
