Auto-Encoding Morph-Tokens for Multimodal LLM
About
For multimodal LLMs, the synergy of visual comprehension (textual output) and generation (visual output) presents an ongoing challenge. This is due to a conflicting objective: for comprehension, an MLLM needs to abstract the visuals; for generation, it needs to preserve the visuals as much as possible. Thus, the objective is a dilemma for visual-tokens. To resolve the conflict, we propose encoding images into morph-tokens to serve a dual purpose: for comprehension, they act as visual prompts instructing MLLM to generate texts; for generation, they take on a different, non-conflicting role as complete visual-tokens for image reconstruction, where the missing visual cues are recovered by the MLLM. Extensive experiments show that morph-tokens can achieve a new SOTA for multimodal comprehension and generation simultaneously. Our project is available at https://github.com/DCDmllm/MorphTokens.
Related benchmarks
| Task | Dataset | Result | Rank | |
|---|---|---|---|---|
| Image Editing | MA5K | LPIPS27.9 | 6 | |
| Image Editing | EVR | CLIP Similarity82.6 | 6 | |
| Image Editing | MagicBrush | CLIPim87.9 | 6 |