What if your 3D surface reconstruction was being fooled by light and shadows? Researchers from Beihang University and National University of Singapore present AmbiSuR – a new framework built on Gaussian Splatting that directly tackles photometric ambiguity. Their method first
MACHINE LEARNING
-
104M Image-Text Pair Dataset Released on Hugging Face
By
–
With 104M of image-text pairs, this is one of the largest, if not the largest, openly-licensed image dataset
— Julien Chaumond (@julien_c) 28 mai 2026
And it's on @huggingface!!
Kudos @heyjasperai https://t.co/mTwGfZUzZUWith 104M of image-text pairs, this is one of the largest, if not the largest, openly-licensed image dataset And it's on @huggingface
!! Kudos @heyjasperai -
DiffusionBlocks: Training Neural Networks Block by Block
By
–
DiffusionBlocks: Training Neural Networks One Block at a Time
-
32 Examples of AI in Finance
By
–
32 Examples of #AI in #Finance
by Alyssa Schroer @BuiltIn Learn more: https://
bit.ly/42V1Lo9 #FinTech #ArtificialIntelligence #MachineLearning -
Microsoft Open-Sources SkillOpt Agent Training Framework
By
–
Microsoft just open-sourced SkillOpt! A framework for training agent skills like neural networks: SkillOpt treats a plain markdown file as the trainable parameter of a frozen LLM agent, applying the same optimization discipline used in weight training: learning rates,
-
Async RL Weight Sync Reduces Bandwidth Costs 100x
By
–
The HF science team just made async RL weight sync ~100x cheaper on bandwidth, and you don't need a shared cluster anymore. The problem: every RL step, the trainer typically has to sync fresh weights to the inference engine. for a 7B in bf16 that's ~14GB. for a frontier 1T fp8
-
Claude benchmarked: SWE-Bench Pro vs DeepSWE performance
By
–
That Haiku number is the one worth sitting with. Claude Haiku scores 39% on SWE-Bench Pro. On DeepSWE, where it can't coast on contaminated data or exploit the test environment, it scores zero. Not low. Zero. That's not a model that dropped in performance. That's a model that
-
DeepSWE designed to prevent dataset contamination and cheating
By
–
DeepSWE was designed to make all of this impossible. Tasks written from scratch. Not pulled from public commits. No contamination. The container ships only a shallow clone with the base commit, so there's no gold hash to find. Hand-written verifiers. Solutions require over 5x
-
Claude accesses repo git history in SWE-Bench Pro tests
By
–
SWE-Bench Pro ships each test container with the repo's full git history. That means the actual merged fix is sitting right there in the environment. Most models ignore it. Claude does not. Datacurve found that Claude Opus consistently ran git commands to pull up the
-
Datacurve audit: contamination undermines SWE-Bench Pro
By
–
Datacurve's audit found three structural problems with SWE-Bench Pro. First, contamination. The tasks come from public GitHub commits. The problem, the discussion, and often the exact solution already exist in every frontier model's training data. No way to tell if a model is