Unverified50% confidenceFactExact time
FDSL (Formula-Driven Supervised Learning) generates images from fractal formulas and discretizes labels to accomplish classification-style pretraining.
1
Sources
50%
Confidence
Long-term
Relevance
10/30/2026
First Seen
Sources
Related Entities
Related Claims
Unverified原生多模态预训练在统一token空间中对图像patch和文本token进行联合自回归训练,使模型在权重层面形成跨模态语义绑定72% similarUnverifiedThe pretraining and fine-tuning paradigm originated from NLP and computer vision, with its core being large-scale data pretraining to learn general low-level features, and downstream tasks adapted using a small amount of labeled data.71% similarUnverifiedTraditional Gaussian proxy rendering methods trained with images bake fixed transfer functions into the model's appearance representation during training, and switching TF presets often requires retraining.71% similarUnverifiedMultimodal classification models tend to recognize states from screen images and make decisions, which is not the same as end-to-end reinforcement learning training.70% similarUnverifiedMultimodal pre-trained models represented by the LayoutLM series jointly model visual, textual, and positional information, advancing the capability frontier in the field of document understanding69% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1114070API
curl https://kongchang.com/api/v1/knowledge/claims/1114070MCP
get_claim(id=1114070)