SolarRenderer is a video renderer that adapts a pretrained MiniMax-H3 33B backbone with two lightweight LoRAs. Structural videos are input into the backbone's native reference pathway at frame positions they describe, without a type identifier, allowing a single control LoRA to follow arbitrary structural representations. The method enables four-step generation, 12.3 times faster than 50-step sampling, while preserving learned controls.
The chart will appear after repeat observations. The current metric comes from the source.
1 source votesReal observations only. History before source connection is not reconstructed.
May be useful for generating realistic videos from structural descriptions, such as game engines or simulators.
MemT-1.0 is a new architecture of memory foundation models that internalizes memory as part of model computation. It uses Gated DeltaNet-2 with a stable token-level memory update mechanism and Mixture-of-Storage to scale native memory. MemT-1.0-Nano and MemT-1.0-Flash models outperformed parametric memory baselines and were competitive with retrieval-augmented methods.
2 days agoDaily PapersWaypoint-OPD uses an intermediate teacher checkpoint to compare early and late shifts in action preferences during policy transfer. It evaluates the consistency between early and late shifts and suppresses the endpoint shift when they conflict.
6 days agoDaily PapersThis paper investigates the extent to which sparse autoencoder (SAE) representations support behavioral compositionality. The study introduces SAE Additivity Residuals (SAR) to measure unexplained behavioral variation above a chosen interaction order. Experiments show that learned SAE coordinates exhibit lower first-order residuals compared to random rotations of the same subspace.
6 days agoDaily PapersWe study how vision-language models (VLMs) can learn to think and actively explore in 3D space. Instead of providing multi-view images, videos, or 3D representations to the model, we preserve the native visual encoding of VLMs, i.e., a single image, one at a time, just like humans, and treat the 3D representation only as an environment with which the model can interact through actions (e.g., move left) and observations (rendered images).
6 days agoDaily PapersThis work introduces Scene-Designer, an agentic framework that learns design intelligence through complementary design and critique experiences. Two datasets, Scene-OTA-35K and Scene-OTC-15K, were created for scene design and quality critique. A Scene-GRPO methodology was proposed to improve design decisions and reduce unnecessary corrections. Experiments demonstrate superior performance in physical validity, functional correctness, and layout coherence.
5 days agoDaily PapersVision-language-action (VLA) models transfer semantic knowledge from vision-language pretraining to robot control, but do not explicitly account for intermediate physical state changes. VLA-Δ uses physical state changes between current and future states as explicit learning targets, with a large-scale pretraining dataset containing annotations of physical state transitions.
6 days ago