Your personalresearch library
Discover papers, track what you're reading, rate and review.Build a library that grows with your curiosity.
Currently Reading
Want to Read
All Papers
High-fidelity human chromosome transfer and elimination
Gianluca Petris, Simona Grazioli, Linda van Bijsterveldt, Pierre Murat, Kim C. Liu, Jakob Birnbaum, Julian E. Sale, Jason W. Chin
Science·2025
<jats:p>The synthesis of human genomes and other gigabase-scale genomes will require new strategies. Here, we realized key steps in our pipeline for building synthetic human chromosomes. We established: (i) the facile transfer of human chromosomes from human cells to mouse embryonic stem cells (assembly cells), where they are haploid, are nonessential, and may be operated on; (ii) the transfer of these human chromosomes from monochromosomal hybrids back into human cells to generate defined, synthetic aneuploidies; and (iii) the elimination of the corresponding endogenous human chromosomes to regenerate diploid cells containing a transferred chromosome. All steps were performed in nontransformed cells without chromothripsis and generated minimal structural variants, insertions, deletions, or single-nucleotide variants.</jats:p>
No ratings yet
View paper →tritonBLAS: Triton-based Analytical Approach for GEMM Kernel Parameter Selection
Ryan Swann, Muhammad Osama, Xiaohu Guo, Bryant Nelson, Lixun Zhang, Alex Brown, Yen Ong, Ali Yazdani, Sean Siddens, Ganesh Dasika, Alex Underwood
arXiv·2025
We present tritonBLAS, a fast and deterministic analytical model that uses architectural parameters like the cache hierarchy, and relative code and data placement to generate performant GPU GEMM kernels. tritonBLAS explicitly models the relationship between architectural topology, matrix shapes, and algorithmic blocking behavior to predict near-optimal configurations without runtime autotuning. Based on this model, we developed and implemented a lightweight GEMM framework entirely within Triton. We evaluate the performance of tritonBLAS across a diverse set of GEMM problem sizes on modern GPUs. tritonBLAS achieves over 95% of the performance of autotuning solutions, while reducing autotuning time to zero. This makes tritonBLAS a practical drop-in replacement for empirical tuning in production HPC and ML workloads.
No ratings yet
View paper →RippleBench: Capturing Ripple Effects Using Existing Knowledge Repositories
Roy Rinberg, Usha Bhalla, Igor Shilov, Flavio P. Calmon, Rohit Gandikota
arXiv·2025
Targeted interventions on language models, such as unlearning, debiasing, or model editing, are a central method for refining model behavior and keeping knowledge up to date. While these interventions aim to modify specific information within models (e.g., removing virology content), their effects often propagate to related but unintended areas (e.g., allergies); these side-effects are commonly referred to as the ripple effect. In this work, we present RippleBench-Maker, an automatic tool for generating Q&A datasets that allow for the measurement of ripple effects in any model-editing task. RippleBench-Maker builds on a Wikipedia-based RAG pipeline (WikiRAG) to generate multiple-choice questions at varying semantic distances from the target concept (e.g., the knowledge being unlearned). Using this framework, we construct RippleBench-Bio, a benchmark derived from the WMDP (Weapons of Mass Destruction Paper) dataset, a common unlearning benchmark. We evaluate eight state-of-the-art unlearning methods and find that all exhibit non-trivial accuracy drops on topics increasingly distant from the unlearned knowledge, each with distinct propagation profiles. To support ongoing research, we release our codebase for on-the-fly ripple evaluation, along with the benchmark, RippleBench-Bio.
★ 4.0 (1)
View paper →