2025-02-03 |
uvos | HIP: add GGML_CUDA_CC_IS_* for amd familys as increasin... |
commit | commitdiff | tree |
2025-02-03 |
Johannes Gäßler | CUDA: use mma PTX instructions for FlashAttention ... |
commit | commitdiff | tree |
2025-02-03 |
Olivier Chafik | `ci`: use sccache on windows instead of ccache (llama... |
commit | commitdiff | tree |
2025-02-03 |
uvos | HIP: require at least HIP 5.5 |
commit | commitdiff | tree |
2025-02-03 |
uvos | HIP: Prepare reduction operators for wave 64 |
commit | commitdiff | tree |
2025-02-03 |
uvos | CUDA/HIP: add warp_size to cuda_device_info |
commit | commitdiff | tree |
2025-02-03 |
Rémy Oudompheng | vulkan: implement initial support for IQ2 and IQ3 quant... |
commit | commitdiff | tree |
2025-02-03 |
Jeff Bolz | vulkan: Catch pipeline creation failure and print an... |
commit | commitdiff | tree |
2025-02-03 |
uvos | HIP: Supress transformation warning in softmax.cu |
commit | commitdiff | tree |
2025-02-03 |
Nikita Sarychev | HIP: Only call rocblas_initialize on rocblas versions... |
commit | commitdiff | tree |
2025-02-03 |
someone13574 | cmake : don't fail on `GGML_CPU=OFF` (llama/11457) |
commit | commitdiff | tree |
2025-02-03 |
Akarshan Biswas | SYCL : SOFTMAX F16 mask support and other fixes (llama... |
commit | commitdiff | tree |
2025-02-03 |
Haus1 | AMD: parse the architecture as supplied by gcnArchName... |
commit | commitdiff | tree |
2025-02-03 |
Ihar Hrachyshka | metal: Handle null returned from MTLCreateSystemDefault... |
commit | commitdiff | tree |
2025-02-03 |
Georgi Gerganov | metal : use residency sets (llama/11427) |
commit | commitdiff | tree |
2025-02-03 |
bandoti | cmake: add ggml find package (llama/11369) |
commit | commitdiff | tree |
2025-02-03 |
Jeff Bolz | vulkan: compile shaders on-demand (llama/11406) |
commit | commitdiff | tree |
2025-02-03 |
uvos | Hip: disable VMM on hip as it seams that it dosent... |
commit | commitdiff | tree |
2025-02-03 |
uvos | hip : Add hipGraph and VMM support to ROCM (llama/11362) |
commit | commitdiff | tree |
2025-02-03 |
Johannes Gäßler | CUDA: fix FP16 cuBLAS GEMM (llama/11396) |
commit | commitdiff | tree |
2025-02-03 |
uvos | rocBLAS: Avoid fp32->fp16->fp32 conversion on cdna... |
commit | commitdiff | tree |
2025-02-03 |
Johannes Gäßler | CPU/CUDA: fix (GQA) mul mat back, add CUDA support... |
commit | commitdiff | tree |
2025-02-03 |
Bernhard M... | cmake : avoid -march=native when reproducible build... |
commit | commitdiff | tree |
2025-02-03 |
amd-dwang | Vulkan-run-test: fix mmq_wg_denoms (llama/11343) |
commit | commitdiff | tree |
2025-02-03 |
Jeff Bolz | vulkan: sort shaders for more deterministic binary... |
commit | commitdiff | tree |
2025-02-03 |
Jeff Bolz | vulkan: fix diag_mask_inf (llama/11323) |
commit | commitdiff | tree |
2025-02-03 |
Radoslav Gerganov | rpc : better caching of the base buffer pointer (llama... |
commit | commitdiff | tree |
2025-02-03 |
Georgi Gerganov | metal : fix out-of-bounds write (llama/11314) |
commit | commitdiff | tree |
2025-02-03 |
Jeff Bolz | vulkan: fix coopmat2 validation failures (llama/11284) |
commit | commitdiff | tree |
2025-02-03 |
Nicolò Scipione | SYCL: Introducing memory host pool (llama/11251) |
commit | commitdiff | tree |
2025-02-03 |
Georgi Gerganov | cmake : add sanitizer flags for llama.cpp (llama/11279) |
commit | commitdiff | tree |
2025-02-03 |
Jeff Bolz | vulkan: fix coopmat2 flash attention for non-contiguous... |
commit | commitdiff | tree |
2025-02-03 |
Radoslav Gerganov | rpc : early register backend devices (llama/11262) |
commit | commitdiff | tree |
2025-02-03 |
Jeff Bolz | vulkan: support copy from f32 to q4_0/q4_1/q5_0/q5_1... |
commit | commitdiff | tree |
2025-02-03 |
Jeff Bolz | vulkan: optimize coopmat2 q4_k/q5_k dequant functions... |
commit | commitdiff | tree |
2025-02-03 |
Jeff Bolz | vulkan: optimize coopmat2 q2_k dequant function (llama... |
commit | commitdiff | tree |
2025-02-03 |
Johannes Gäßler | CUDA: backwards pass for misc. ops, add tests (llama... |
commit | commitdiff | tree |
2025-02-03 |
fj-y-saito | ggml: aarch64: implement SVE kernels for q4_K_q8_K... |
commit | commitdiff | tree |
2025-02-03 |
Eve | vulkan: scale caching for k quants + misc fixes (llama... |
commit | commitdiff | tree |
2025-02-03 |
Junil Kim | fix: ggml: fix vulkan-shaders-gen build (llama/10448) |
commit | commitdiff | tree |
2025-02-03 |
Johannes Gäßler | RoPE: fix back, CUDA support for back + noncont. (llama... |
commit | commitdiff | tree |
2025-02-03 |
Akarshan Biswas | SYCL: Add gated linear attention kernel (llama/11175) |
commit | commitdiff | tree |
2025-02-03 |
William Tambellini | ggml : add option to not print stack on abort (ggml... |
commit | commitdiff | tree |
2025-02-03 |
issixx | ggml-cpu : fix ggml_graph_compute_thread did not termin... |
commit | commitdiff | tree |
2025-02-03 |
Georgi Gerganov | ci : dummy commit to trigger CI |
commit | commitdiff | tree |
2025-01-21 |
KITAITI Makoto | ruby : Make context accept initial parameters, API... upstream/1.7.4+33 |
commit | commitdiff | tree |
2025-01-18 |
Corey Earwood | whisper.objc : fix build and CI |
commit | commitdiff | tree |
2025-01-14 |
Georgi Gerganov | talk-llama : sync llama.cpp |
commit | commitdiff | tree |
2025-01-14 |
Georgi Gerganov | sync : ggml |
commit | commitdiff | tree |
2025-01-14 |
Johannes Gäßler | GGUF: C++ refactor, backend support, misc fixes (skip... |
commit | commitdiff | tree |
2025-01-14 |
lhez | ggml : add opencl backend (skip) (llama/10693) |
commit | commitdiff | tree |
2025-01-14 |
Andreas Kieslinger | cuda : CUDA Graph Compute Function Refactor (precursor... |
commit | commitdiff | tree |
2025-01-14 |
Radoslav Gerganov | ggml : do not define GGML_USE_CUDA when building with... |
commit | commitdiff | tree |
2025-01-14 |
0cc4m | Vulkan: Fix float16 use on devices without float16... |
commit | commitdiff | tree |
2025-01-14 |
Molly Sophia | llama: add support for QRWKV6 model architecture (llama... |
commit | commitdiff | tree |
2025-01-14 |
Akarshan Biswas | SYCL: Refactor ggml_sycl_compute_forward (llama/11121) |
commit | commitdiff | tree |
2025-01-14 |
hydai | fix: add missing msg in static_assert (llama/11143) |
commit | commitdiff | tree |
2025-01-14 |
amritahs-ibm | llamafile : ppc64le MMA INT8 implementation (llama... |
commit | commitdiff | tree |
2025-01-14 |
Mathieu Baudier | Disable GL_KHR_cooperative_matrix Vulkan extension... |
commit | commitdiff | tree |
2025-01-14 |
ag2s20150909 | fix: Vulkan shader gen binary path when Cross-compiling... |
commit | commitdiff | tree |
2025-01-14 |
Johannes Gäßler | GGUF: C++ refactor, backend support, misc fixes (llama... |
commit | commitdiff | tree |
2025-01-14 |
Diego Devesa | ggml-backend : only offload from host buffers (fix... |
commit | commitdiff | tree |
2025-01-14 |
Diego Devesa | ggml-backend : only offload from host buffers (llama... |
commit | commitdiff | tree |
2025-01-14 |
Radoslav Gerganov | rpc : code cleanup (llama/11107) |
commit | commitdiff | tree |
2025-01-14 |
Akarshan Biswas | SYCL: Use get_multi_ptr instead of deprecated get_point... |
commit | commitdiff | tree |
2025-01-14 |
Johannes Gäßler | CUDA: add BF16 support (llama/11093) |
commit | commitdiff | tree |
2025-01-14 |
0cc4m | Vulkan: Add device-specific blacklist for coopmat for... |
commit | commitdiff | tree |
2025-01-14 |
matt23654 | Support for models with non-512-aligned tensors over... |
commit | commitdiff | tree |
2025-01-14 |
Gilad S. | fix: Vulkan shader gen binary path (llama/11037) |
commit | commitdiff | tree |
2025-01-14 |
Radoslav Gerganov | ggml : allow loading backend with env variable (ggml... |
commit | commitdiff | tree |
2025-01-14 |
Georgi Gerganov | scripts : sync opencl, gguf |
commit | commitdiff | tree |
2025-01-13 |
Georgi Gerganov | whisper : fix gpu device selection (#2728) |
commit | commitdiff | tree |
2025-01-13 |
Georgi Gerganov | server : fix build (#2718) |
commit | commitdiff | tree |
2025-01-13 |
Georgi Gerganov | talk-llama : sync llama.cpp (#2709) |
commit | commitdiff | tree |
2025-01-13 |
NETZkultur... | server : generate unique tmp filenames (#2718) |
commit | commitdiff | tree |
2025-01-09 |
Sandro Hanea | whisper : add whisper_full_get_segment_no_speech_prob_f... |
commit | commitdiff | tree |
2025-01-07 |
Jayant | readme : add docker instructions (#2711) |
commit | commitdiff | tree |
2025-01-06 |
Adam Jones | docs: Fix main -> whisper-cli in download scripts ... |
commit | commitdiff | tree |
2025-01-06 |
Georgi Gerganov | release : v1.7.4 upstream/1.7.4 |
commit | commitdiff | tree |
2025-01-06 |
Georgi Gerganov | ci : cont |
commit | commitdiff | tree |
2025-01-06 |
Georgi Gerganov | ci : fix ubuntu runner names |
commit | commitdiff | tree |
2025-01-04 |
Yusuf Redžić | cli : fix segfault on missing argument (#2700) |
commit | commitdiff | tree |
2025-01-04 |
Georgi Gerganov | ci : fix arm builds |
commit | commitdiff | tree |
2025-01-04 |
Georgi Gerganov | sync : ggml |
commit | commitdiff | tree |
2025-01-04 |
Georgi Gerganov | ggml : do not install metal source when embed library... |
commit | commitdiff | tree |
2025-01-04 |
Georgi Gerganov | metal : avoid uint (llama/11019) |
commit | commitdiff | tree |
2025-01-04 |
Srihari-mcw | ggml : fixes for AVXVNNI instruction set with MSVC... |
commit | commitdiff | tree |
2025-01-04 |
Jeff Bolz | vulkan: optimize mul_mat for small values of N (llama... |
commit | commitdiff | tree |
2025-01-04 |
Jeff Bolz | vulkan: im2col and matmul optimizations for stable... |
commit | commitdiff | tree |
2025-01-04 |
Jeff Bolz | vulkan: Use push constant offset to handle misaligned... |
commit | commitdiff | tree |
2025-01-04 |
Eve | vulkan: multi-row k quants (llama/10846) |
commit | commitdiff | tree |
2025-01-04 |
Peter | examples, ggml : fix GCC compiler warnings (llama/10983) |
commit | commitdiff | tree |
2025-01-04 |
Djip007 | ggml : more perfo with llamafile tinyblas on x86_64... |
commit | commitdiff | tree |
2025-01-04 |
Diego Devesa | ggml : use wstring for backend search paths (llama... |
commit | commitdiff | tree |
2025-01-04 |
Diego Devesa | ggml : fix arm enabled features check (llama/10961) |
commit | commitdiff | tree |
2025-01-04 |
Diego Devesa | ggml : fix const usage in SSE path (llama/10962) |
commit | commitdiff | tree |
2025-01-04 |
yuri@FreeBSD | ggml : fix run-time on FreeBSD in get_executable_path... |
commit | commitdiff | tree |
2025-01-04 |
Jeff Bolz | vulkan: build fixes for 32b (llama/10927) |
commit | commitdiff | tree |
2025-01-04 |
Jeff Bolz | vulkan: optimize coopmat2 dequant functions (llama... |
commit | commitdiff | tree |
2025-01-04 |
Adrien Gallouët | ggml-cpu: replace NEON asm with intrinsics in ggml_gemv... |
commit | commitdiff | tree |
next |