AI Resonance · 阅读最新日报 · AI 入门推荐

2026-11-15 · America/Los_Angeles · 社区动态 · #2

[Meta FAIR + NYU] Unbundling Multimodal Vision Encoding: training 200+ MLLMs to find encoder pretraining is optional but vision-only parameters are not [R]

X threads (with videos!): https://x.com/choang333/status/2121335484236386574 https://x.com/_ellisbrown/status/2121335477244428419 Paper: https://www.alphaxiv.org/abs/2611.17334 Blog and FAQ: https://unbundling.github.io Labs disagree on vision encoders right now. Gemma 4 12B dropped the encoder, Kimi K3 trains its encoder from scratch, and Inkling 2 Small brought an encoder back. Published models can't settle this, because they differ in data and compute and rarely account for encoder pretraining. A standard encoder also bundles three choices: separate pretraining, parameters that only…

[Meta FAIR + NYU] Unbundling Multimodal Vision Encoding: training 200+ MLLMs to find encoder pretraining is optional but vision-only parameters are not [R]

热度 49.0 / 100;排名与评分保留该期记录。