“I'm still not sure why they keep LeCun at Facebook; his single most-cited contribution to the field in 2024 has been with NYU[0], not Facebook. What is his role at Facebook exactly, has he explained it? I recall him making all the wrong predictions in 2023 what's changed? Chollet is similarly a mystery to me; it feels like these guys were busy riffing CNN…”
Cambrian-1: A Fully Open, Vision-Centric Exploration of Multimodal LLMs
We introduce Cambrian-1, a family of multimodal LLMs (MLLMs) designed with a vision-centric approach. While stronger language models can enhance multimodal capabilities, the design choices for vision components are often insufficiently explored and disconnected from visual representation learning research. This gap hinders accurate sensory grounding in real-world scenarios. Our study uses LLMs and visual instruction tuning as an interface to evaluate various visual representations, offering new insights into different models and architectures -- self-supervised, strongly supervised, or combinations thereof -- based on experiments with over 20 vision encoders. We critically examine existing MLLM benchmarks, address the difficulties involved in consolidating and interpreting results from various tasks, and introduce a new vision-centric benchmark, CV-Bench. To further improve visual grounding, we propose the Spatial Vision Aggregator (SVA), a dynamic and spatially-aware connector that integrates high-resolution vision features with LLMs while reducing the number of tokens. Additionally, we discuss the curation of high-quality visual instruction-tuning data from publicly available sources, emphasizing the importance of data source balancing and distribution ratio. Collectively, Cambrian-1 not only achieves state-of-the-art performance but also serves as a comprehensive, open cookbook for instruction-tuned MLLMs. We provide model weights, code, supporting tools, datasets, and detailed instruction-tuning and evaluation recipes. We hope our release will inspire and accelerate advancements in multimodal systems and visual representation learning.
候选层:至少 2 个不同的人、在 2 个不同月份提到过,还没到主书架的 3 人 3 个月。没有做立场标注,所以只写“被提及”。
HN 书评
推荐在前,同类按时间 · 每篇最多保留 20 条,统计用全部提及“A lot of problems jump out to me with this article, particularly with the explanation of multi-modal LLMs. I'll say that I _do_ agree with the thrust of the article. Don't trust LLMs. But they probably should have argued legitimate issues with VLM based OCR, rather than try to talk about how VLMs are somehow fundamentally flawed or something. > LLMs proc…”