GitHub 新星
10 条
arXiv 论文
10 条
-
1
One Figure, Every Canvas: Editable Flowchart Relayout via Agentic Pipeline
Pipeline figures in ML papers must be repurposed across many canvases, including paper columns, 16:9 slides, portrait posters, 1:1 social teasers, 9:16 phone previews. Each format imposes a differentShih-Chen Tseng, Chih-Hsuan Chen, Ryan Yang
-
2
Base Models Can Reason By Taking a Cue From Training Data
In this paper, we study how training data creates associations between the tokens at the start of a base model's response and the reasoning behavior that follows. First, we demonstrate that fixing parSophie L. Wang, Amil Dravid, Rulin Shao
-
3
BiasFlow: Geometric Monitoring and Backbone Regularization for Spurious Feature Reliance
Worst-group accuracy (WGA) evaluates a trained predictor but does not characterize how its frozen backbone behaves when a new head is learned. We introduce BiasFlow, a hook-based toolkit for monitorinHaojin Deng, Zhiping Lin, Yimin Yang
-
4
Learning to Read the Contextual Tokens in Diffusion Transformers
Multimodal Diffusion Transformers (MM-DiTs) jointly process visual and textual representations throughout generation. These models repeatedly update the text tokens through multimodal attention, formiOmer Dahary, Etai Sella, Hadar Averbuch-Elor
-
5
Recursive Video In-Context Learning for Agentic Robot
LLM agents that orchestrate frozen vision-language-action (VLA) policies improve across episodes through text memory, which records what the agent did but not how the task is done. A demonstration vidWenrui Bao, Xinxin Liu, Bingxin Xu
-
6
Direct Intermediate Initialization for Tilted Diffusion Samplers
Some diffusion posterior samplers construct Gaussian-tilted intermediate distributions along the reverse process. We observe that these targets can be pulled back to clean-space posteriors with weakerGregory D. Bellchambers
-
7
Towards Looped Models Done Right, Part II: Rethinking at Fixed Points
Every recurrence of a looped language model adds cost in training, decoding, prefill, and reinforcement learning (RL). The closer recurrent states get to fixed points, the less the path to them matterBenhao Huang, Chufan Shi, Junlin Chen
-
8
UniSlider: Perceptually Uniform Sliders for Continuous Image Editing
Sliders provide an intuitive interface for continuous image editing. In current generative approaches, however, the slider is simply a rescaling of the method's strength parameter, such as an adapterDavid Serrano-Lozano, Duygu Ceylan, Yannick Hold-Geoffroy
-
9
MemPilot: Orchestrating On-Demand Multimodal Memory Curation for LLM Agents
Memory has become integral to the LLM agent ecosystem, supporting information retention and reuse across interactions. However, most existing agent memory systems construct memory in a query-agnosticHaozhen Zhang, Haodong Yue, Quanyu Long
-
10
CLIFT: Conformal Self-Verification for Web Agent Training and Test-Time Scaling
Open-source web agents are now strong enough to execute realistic browser tasks, but training them with reinforcement learning still depends on weak supervision: binary task success is too sparse forYifan Zhang, Yutong Dai, Viraj Prabhu
Hacker News
10 条
- 1 Mistral Large 4
- 2 Nobel Prize in Physics 2026: Francis Halzen
- 3 Sharing AI progress in mathematics
- 4 Gleam doesn't compile to Erlang source anymore
- 5 OpenTPU – An open-source AI accelerator, developed by AI
- 6 EmbeddingGemma 2: An open, lightweight multimodal embedding model
- 7 Paramount Skydance has completed its $111B merger with Warner Bros. Discovery
- 8 What's Earth's dominant species by mass?
- 9 Benchmark in Milliseconds
- 10 Decisions API is in public beta