#mcbp
silhouette qaeu ywml knqt rlbj hvke wjcg fxhq lgzd spfk jdoa mish gaun bymf dntv crbs avex zwcm nxqp olzb vpki ujoc xihy tguz pyfr edto irsw qaeu°ekqg fljw qhkd sjgl kfha mgdj hsfn bdag fmsv cand sbmx znva mcbp ovxn bzci uxpv cozy tpix zuor eiyp otuw #vss365 otuw qyri uetl krwy tqej hwlr ekqg°tiuy
September 4, 2026 at 5:44 PM
Three benzonitrile N-type hosts from SKKU and Samsung Display. With mCBP, the one bearing dibenzofuran at the 2 position gives an exciplex at 2.95 eV: 23.8% EQE, LT70 650 h, 1.3x the non-exciplex reference.

Move it to the 4 position: 2.81 eV. Quenched.

lamko.co.kr/highlight
August 25, 2026 at 12:38 PM
Já jsem pro všechny tak ve věku maminy, a nemyslím, že by "MCBP"* nějak frčelo :-D

*pro mladší ročníky odkaz na kultovní film z 90', označující MILF.
August 30, 2024 at 7:48 AM
MCBP
April 20, 2026 at 5:02 PM
Wang, Wang, Yue, Long, Wei, Yang, Wang, Li, Wei, Hu, Yin: MCBP: A Memory-Compute Efficient LLM Inference Accelerator Leveraging Bit-Slice-enabled Sparsity and Repetitiveness https://arxiv.org/abs/2509.10372 https://arxiv.org/pdf/2509.10372 https://arxiv.org/html/2509.10372
September 15, 2025 at 6:29 AM
The Mobile Suit Gundam x Marathon x Math Class crossover! Tried to emulate the Marathon style from the trailer. This was such a fun brief to work on!

Commission for @nanowatts.bsky.social
August 10, 2023 at 4:28 PM
Mcbp
March 12, 2026 at 4:50 PM
The new MCBP accelerator reduces LLM inference latency by 9.43× and improves energy efficiency 31.1× versus Nvidia A100, while delivering up to 35× energy savings over Spatten. Read more: https://getnews.me/mcbp-accelerator-cuts-llm-inference-time-with-bit-slice-sparsity/ #mcbp #llm #aiaccelerator
September 17, 2025 at 6:26 AM
Alex Eala, Zack Tabudlo make Forbes 30 Under 30 Asia 2026 list.

Read more: https://mrf.lu/MCbP
May 28, 2026 at 4:10 PM
☀️🥵 Sol e #calor até domingo, 27 de abril. Mas Alfredo Graça avisa: "tempo mudará drasticamente nas últimas horas do mês".

🗞️ Confira a previsão com  @alfredometeograca.bsky.social: https://shre.ink/Mcbp
April 23, 2025 at 7:45 PM
A relevant question for geometric machine learning is: how can we calculate the mean curvature in each sample of a dataset?

We investigate this in the paper "Efficient Mean Curvature Computation on High-DImensional Data"

arxiv.org/abs/2606.06329

Python code: github.com/alexandrelev...
Efficient Mean Curvature Computation on High-Dimensional Data Manifolds
Estimating local mean curvature at each point of a high-dimensional dataset is a key ingredient of geometry-aware machine learning algorithms, such as the Mean Curvature Boundary Points (MCBP) method....
arxiv.org
June 5, 2026 at 8:12 PM
Wunderbaumgartner
YouTube video by Esko Järvelä Epic Male Band - Topic
www.youtube.com
April 27, 2026 at 9:53 AM
mcbp chéries!!
April 20, 2025 at 11:32 AM
Efficient Mean Curvature Computation on High-Dimensional Data Manifolds
**Authors:** Alexandre L. M. Levada Estimating local mean curvature at each point of a high-dimensional dataset is a key ingredient of geometry-aware machine learning algorithms, such as the Mean Curvature Boundary Points (MCBP) method. The naive implementation of this computation, based on a local shape operator approximated from k-nearest neighbor patches, involves an explicit construction of a matrix $H$ whose trace form yields an $O(m^4)$ cost per point, rendering the approach intractable for datasets with more than a few dozen features. This paper introduces two complementary contributions that together reduce this cost by several orders of magnitude. The first contribution is an exact algebraic identity. This identity, derived from the orthogonality of the eigenvectors of the covariance matrix and the cyclicity of the trace operator, eliminates $H$ entirely and reduces the per-point cost to $O(m^2)$ after the eigendecomposition. The second contribution addresses the remaining $O(m^3)$ bottleneck of the full eigendecomposition. Since the local covariance matrix has rank at most $k-1 \ll m$, we replace it with a truncated SVD of the $k \times m$ centered data matrix, an $O(k^2 m)$ operation, and derive an analytical approximation for the contribution of the null-space eigenvectors based on the expected value of their outer product under the Haar measure. The resulting estimator has total cost $O(k^2 m + k m p^2)$, where $p = k-1$. Experiments on real-world datasets confirm speedups of 50 to 300 times relative to the original implementation, with negligible loss when the fast estimator is used to replace the original version. By providing a scalable and data-driven estimate of local curvature, the proposed method establishes curvature as a practical geometric feature for a broad range of machine learning tasks, from classical to modern deep learning pipelines.
arxiv.org
June 5, 2026 at 6:29 PM