ggml : adapt AMX to tensor->grad removal (llama/0)

ggml : fix compile warnings (llama/0)

llamafile : fix include path (llama/0)

vulkan: Optimize some mat-vec mul quant shaders (llama/10296)

ggml : optimize Q4_0 into Q4_0_X_Y repack (llama/10324)

Make updates to fix issues with clang-cl builds while using AVX512 flags (llama/10314)

ggml: new optimization interface (ggml/988)

ggml : remove duplicated sources from the last sync (ggml/1017)

ggml : fix some build issues

sync : leftovers (ggml/0)

cmake : restore CMakeLists.txt (llama/10256)

AVX BF16 and single scale quant optimizations (llama/10212)

sycl: Use syclcompat::dp4a (llama/10267)

backend cpu: add online flow for aarch64 Q4_0 GEMV/GEMM kernels (llama/9921)

ggml : build backends as libraries (llama/10256)

scripts : update sync

release : v1.7.2

sycl: fix example build (#2570)

ci : use local ggml in Android build (#2567)

ggml : tmp workaround for whisper.cpp (skip) (#2565)

update : readme

scripts : fix sync path

whisper.swiftui : switch Mac dest to Mac (Designed for iPad) (#2562)

cmake : fix ppc64 check (#0)

whisper : include ggml-cpu.h (#0)

build : fixes

talk-llama : sync llama.cpp

whisper : fix build (#0)

sync : ggml

sycl : Fixes to broken builds and test-backend-ops (llama/10257)

vulkan: Optimize contiguous copies (llama/10254)

vulkan: Throttle the number of shader compiles during the build step. (llama/10222)

metal : more precise Q*K in FA vec kernel (llama/10247)

vulkan: Fix newly added tests for permuted mul_mat and 1D im2col (llama/10226)

metal : reorder write loop in mul mat kernel + style (llama/10231)

metal : fix build and some more comments (llama/10229)

metal : fix F32 accumulation in FA vec kernel (llama/10232)

metal : hide debug messages from normal log

ggml: fix zero division in ‘dne’ calculation in CUDA COUNT_EQUAL operator when ‘ne’ is small (#10213)

ggml : optimize llamafile cpu matrix multiplication for ppc64le (llama/10156)

metal : opt-in compile flag for BF16 (llama/10218)

metal : improve clarity (minor) (llama/10171)

metal : optimize FA kernels (llama/10171)

ggml : add ggml-cpu.h to the public headers (llama/10204)

fix q4_0_8_8 format for corrupted tokens issue (llama/10198)

Optimize RWKV6 Operator Naming and Implement Multi-core CPU/ SYCL Acceleration (llama/10133)

metal : add BF16 support (llama/8439)

metal : fix from ptr buffer name (llama/10189)

ggml : adjust is_first_call init value (llama/10193)

metal : add quantized FA support (llama/10149)