NeoMME: 260M/800M multimodal-native multilingual encoders drop the vision tower
antoine_chaffin · x · 2026-09-04
Researcher tonywu71 introduces NeoMME, a family of 260M and 800M multimodal-native, multilingual efficient encoders. A single bidirectional Transformer processes both text tokens and raw image patches — with no pretrained vision tower, text encoder, or decoder. NLP researcher bclavie and others amplified it, calling it impressive, thorough work that outperforms what they'd have designed for image encoders.
More from Multimodal
- FastVideo open-sources MiniMax FastH3 Omni with training code, 12x faster on Blackwell — multimodalart · 2026-10-21
- Megalith 008: stunning AI-generated colossal architecture series — azed_ai · 2026-10-21
- Higgsfield offers $25K prizes for open-source ComfyUI rebuilds of its apps — MuziqueComfyUI · 2026-10-21
- Stunning AI-generated elderly portrait blurs the line with real photography — LudovicCreator · 2026-10-21
- Writer with no drawing skills uses AI to publish third episode of original manga — OriginalMangaStudio · 2026-10-21
- HeyGen's Professional Voice Clone needs just 30 min of audio to sound exactly like you — rohanpaul_ai · 2026-10-21