Inbox triage 2026-09-19
Social and low-evidence quarantine moved to raw as rejected. 168 high-specificity articles remain in inbox; four sources were filed.
Knowledge catalogue
Social and low-evidence quarantine moved to raw as rejected. 168 high-specificity articles remain in inbox; four sources were filed.
llama.cpp release b9589 is a CUDA maintenance update that addresses data-race conditions in the ssm_scan_f32 kernel function by adding missing synchronization barriers for shared memory reuse. The rel
Release b8987 of llama.cpp updates cpp-httplib to version 0.43.2 . This is a minor maintenance release from the llama.cpp project, which provides C/C++ inference for large language models. The release
Release b8861 of llama.cpp removed /api endpoints from the server, including the /api/tags endpoint. The release was published on April 20, 2026. This is a maintenance update to the llama.cpp project,
llama.cpp release **b8737** is a focused maintenance build that adds missing CUDA error handling to the ggml backend. Specifically, it checks the return values of NVIDIA CUB library calls used in t...
llama.cpp is an open-source C/C++ project for LLM inference , and build b9301 is a specific release version in the project's development history. This build number represents an incremental developmen
b9240 is a release of llama.cpp that includes a fix for the --help option related to the --verbosity flag . The release provides prebuilt binaries across multiple platforms including macOS (Apple Sili