Pre-release / continuous build: 실험적 빌드이며 안정 버전 릴리스가 아닙니다.
sampling : use greedy selection for eligible temperature-zero chains (#29797) * sampling : use greedy selection for eligible temperature-zero chains Assisted-by: OpenAI Codex * Apply suggestion from @ggerganov Co-authored-by: Georgi Gerganov * sampling: simplify zero-temperature greedy eligibility CPU와 grammar/reasoning-budget 경로에서 동일한 greedy 선택을 허용합니다. 동적 temperature 및 요청된 확률에는 분포 샘플링을 유지합니다. 기존 sampler 테스트에서 공통 sampler 선택과 확률 동작을 다룹니다. Assisted-by: OpenAI Codex * sampling : use greedy selection after final top-k with k=1 Assisted-by: OpenAI Codex * cont : clean-up --------- Co-authored-by: Georgi GerganovWebsite: - https://llama.app
Attestations: - https://github.com/ggml-org/llama.cpp/attestations/53574365
macOS/iOS: - macOS Apple Silicon (arm64) - macOS Apple Silicon (arm64, KleidiAI 활성화) DISABLED - macOS Intel (x64) - iOS XCFramework
Linux: - Ubuntu x64 (CPU) - Ubuntu arm64 (CPU) - Ubuntu s390x (CPU) - Ubuntu x64 (Vulkan) - Ubuntu arm64 (Vulkan) - Ubuntu x64 (CUDA 12) - CUDA 12.8 라이브러리 - Ubuntu x64 (CUDA 13) - CUDA 13.4 라이브러리 - Ubuntu arm64 (CUDA 13) - CUDA 13.4 라이브러리 - Ubuntu x64 (ROCm 10.0) - Ubuntu x64 (OpenVINO) - Ubuntu x64 (SYCL FP32) - Ubuntu x64 (SYCL FP16) - Linux arm64 (Snapdragon: CPU, Adreno GPU, Hexagon NPU) - 설정 가이드
Android: - Android arm64 (CPU) - Android arm64 (Snapdragon: CPU, Adreno GPU, Hexagon NPU) - 설정 가이드
윈도우(Windows): - 윈도우(Windows) x64 (CPU) - 윈도우(Windows) arm64 (CPU) - 윈도우(Windows) arm64 (OpenCL Adreno) - 윈도우(Windows) x64 (CUDA 12) - CUDA 12.4 DLL들 - 윈도우(Windows) x64 (CUDA 13) - CUDA 13.4 DLL들 - 윈도우(Windows) arm64 (CUDA 13) - CUDA 13.4 DLL들 - 윈도우(Windows) x64 (Vulkan) - 윈도우(Windows) arm64 (Vulkan) - 윈도우(Windows) x64 (OpenVINO) - 윈도우(Windows) x64 (SYCL) - 윈도우(Windows) x64 (ROCm 10.0)
openEuler: - DISABLED - openEuler x86 (310p) 환경 - openEuler x86 (910b, ACL Graph) 환경 - openEuler aarch64 (310p) 환경 - openEuler aarch64 (910b, ACL Graph) 환경
UI: - UI