/LLaVolta

[NeurIPS 2024] Efficient Multi-modal Models via Stage-wise Visual Context Compression

Primary LanguagePythonApache License 2.0Apache-2.0

Watchers