ci : fix building workflow for linux/arm64 container (#2555)

ruby : extend API (#2551)

whisper.swiftui : add model download list & bench methods (#2546)

ruby : fix the instructions (#2548)

ggml : vulkan logs (#2547)

examples : fix ffmpeg v5 build (#2543)

whisper : fix extra memory usage (#2534)

whisper : backend registry init before model load

talk-llama : sync llama.cpp

sync : ggml

ggml : add AMX backend (llama/8998)

metal : support permuted matrix multiplicaions (llama/10033)

CUDA: fix insufficient buffer clearing for MMQ (llama/10032)

CUDA: fix MMQ for non-contiguous src0, add tests (llama/10021)

increase cuda_cpy block size (ggml/996)

metal : add POOL2D and fix IM2COL (llama/9943)

Adapt to dynamically loadable backends mechanism (llama/9970)

ggml : add asserts for type conversion in fattn kernels (llama/9971)

rpc : pack only RPC structs (llama/9959)

fix mul_mat_vec_q and *_vec_q error (llama/9939)

rpc : backend refactoring (llama/9912)

Add SYCL Backend registry, device and Event Interfaces (llama/9705)

add amx kernel for gemm (llama/8998)

vulkan : add backend registry / device interfaces (llama/9721)

fix: allocating CPU buffer with size `0` (llama/9917)

fix: use `vm_allocate` to allocate CPU backend buffer on macOS (llama/9875)

CUDA: fix 1D im2col, add tests (ggml/993)

Fix cann compilation error (llama/9891)

Vectorize load instructions in dmmv f16 CUDA kernel (llama/9816)

ggml : move more prints to the ggml log system (llama/9839)

rpc : add backend registry / device interfaces (llama/9812)

musa: add docker image support (llama/9685)

ggml : fix BLAS with unsupported types (llama/9775)

ggml : add backend registry / device interfaces to BLAS backend (llama/9752)

Update building for Android (llama/9672)

ggml : add metal backend registry / device (llama/9713)

metal : single allocation of encode_async block (llama/9747)

ggml-alloc : remove buffer_id from leaf_alloc (ggml/987)

scripts : sync amx

ggml : alloc ggml_contexts on the heap (#2525)

ci : fix openblas build (#2511)

scripts : add turbo-q8_0 to the benchmark

whisper : minor compile warning

whisper : move new-segment callback after DTW step (#2515)

ruby : fix installation test (#2519)

ruby : add more APIs (#2518)

ruby : support new-segment callback (#2506)

ruby : add Metal support (#2516)

whisper : fix index overflow in token-level timestamp logic (#2505)

readme : update links and make commands (#2489)