Commit Graph

  • 80fa576254 vulkan: implement initial support for IQ2 and IQ3 quantizations (llama/11360) Rémy Oudompheng 2025-01-29 18:29:39 +01:00
  • 75e7d0585e vulkan: Catch pipeline creation failure and print an error message (llama/11436) Jeff Bolz 2025-01-29 09:26:50 -06:00
  • 682a6f5f87 HIP: Supress transformation warning in softmax.cu uvos 2025-01-28 23:06:32 +01:00
  • 115716d109 HIP: Only call rocblas_initialize on rocblas versions with the multiple instantation bug (llama/11080) Nikita Sarychev 2025-01-28 07:42:20 -08:00
  • b2cfef655b cmake : don't fail on GGML_CPU=OFF (llama/11457) someone13574 2025-01-28 09:15:34 -05:00
  • 22e3df0afa SYCL : SOFTMAX F16 mask support and other fixes (llama/11261) Akarshan Biswas 2025-01-28 15:26:58 +05:30
  • 028511d349 AMD: parse the architecture as supplied by gcnArchName (llama/11244) Haus1 2025-01-27 08:58:17 -05:00
  • 70c4038842 metal: Handle null returned from MTLCreateSystemDefaultDevice() (llama/11441) Ihar Hrachyshka 2025-01-27 02:41:59 -05:00
  • 8639c003a9 metal : use residency sets (llama/11427) Georgi Gerganov 2025-01-26 20:06:16 +02:00
  • d5d831da65 cmake: add ggml find package (llama/11369) bandoti 2025-01-26 12:07:48 -04:00
  • 7230a6e1c8 vulkan: compile shaders on-demand (llama/11406) Jeff Bolz 2025-01-25 15:29:57 -06:00
  • a160fa0f3a Hip: disable VMM on hip as it seams that it dosent work in some configurations (llama/11420) uvos 2025-01-25 21:01:12 +01:00
  • 0282ad8fd1 hip : Add hipGraph and VMM support to ROCM (llama/11362) uvos 2025-01-25 00:02:23 +01:00
  • 9e467815d4 CUDA: fix FP16 cuBLAS GEMM (llama/11396) Johannes Gäßler 2025-01-24 21:02:43 +01:00
  • 727891d9bf rocBLAS: Avoid fp32->fp16->fp32 conversion on cdna (llama/11356) uvos 2025-01-24 17:50:49 +01:00
  • c262dc80e2 CPU/CUDA: fix (GQA) mul mat back, add CUDA support (llama/11380) Johannes Gäßler 2025-01-24 12:38:31 +01:00
  • 30767b4c4e cmake : avoid -march=native when reproducible build is wanted (llama/11366) Bernhard M. Wiedemann 2025-01-24 12:21:35 +01:00
  • 16eeb31933 Vulkan-run-test: fix mmq_wg_denoms (llama/11343) amd-dwang 2025-01-23 15:14:28 +08:00
  • ba523d5e22 vulkan: sort shaders for more deterministic binary (llama/11315) Jeff Bolz 2025-01-23 01:07:50 -06:00
  • 3736706139 vulkan: fix diag_mask_inf (llama/11323) Jeff Bolz 2025-01-23 01:01:17 -06:00
  • 58640aa456 rpc : better caching of the base buffer pointer (llama/11331) Radoslav Gerganov 2025-01-21 15:06:41 +02:00
  • 5183a05e56 metal : fix out-of-bounds write (llama/11314) Georgi Gerganov 2025-01-21 08:48:13 +02:00
  • 0dcada42d4 vulkan: fix coopmat2 validation failures (llama/11284) Jeff Bolz 2025-01-20 10:38:32 -06:00
  • d507b4cebe SYCL: Introducing memory host pool (llama/11251) Nicolò Scipione 2025-01-19 14:33:34 +01:00
  • 90171055f3 cmake : add sanitizer flags for llama.cpp (llama/11279) Georgi Gerganov 2025-01-18 16:18:15 +02:00
  • 668306ff2b vulkan: fix coopmat2 flash attention for non-contiguous inputs (llama/11281) Jeff Bolz 2025-01-18 02:26:50 -06:00
  • fdc21fc87b rpc : early register backend devices (llama/11262) Radoslav Gerganov 2025-01-17 10:57:09 +02:00
  • 7183a1eb72 vulkan: support copy from f32 to q4_0/q4_1/q5_0/q5_1/q8_0/iq4_nl (llama/11166) Jeff Bolz 2025-01-16 15:47:10 -06:00
  • 09f3c66648 vulkan: optimize coopmat2 q4_k/q5_k dequant functions. (llama/11206) Jeff Bolz 2025-01-16 15:23:49 -06:00
  • 62e2414620 vulkan: optimize coopmat2 q2_k dequant function (llama/11130) Jeff Bolz 2025-01-16 15:16:39 -06:00
  • de49024e49 CUDA: backwards pass for misc. ops, add tests (llama/11257) Johannes Gäßler 2025-01-16 16:43:38 +01:00
  • db6383094c ggml: aarch64: implement SVE kernels for q4_K_q8_K vector dot (llama/11227) fj-y-saito 2025-01-16 18:11:49 +09:00
  • 164f13c6a9 vulkan: scale caching for k quants + misc fixes (llama/11081) Eve 2025-01-15 19:50:13 +00:00
  • 02aa86230a fix: ggml: fix vulkan-shaders-gen build (llama/10448) Junil Kim 2025-01-15 22:17:42 +09:00
  • 54a2ee648f RoPE: fix back, CUDA support for back + noncont. (llama/11240) Johannes Gäßler 2025-01-15 12:51:37 +01:00
  • 9700cfb0a3 SYCL: Add gated linear attention kernel (llama/11175) Akarshan Biswas 2025-01-15 08:50:17 +05:30
  • 8e0143e205 ggml : add option to not print stack on abort (ggml/1081) William Tambellini 2025-01-23 11:59:08 -08:00
  • f12559d590 ggml-cpu : fix ggml_graph_compute_thread did not terminate on abort. (ggml/1065) issixx 2025-01-17 21:29:08 +09:00
  • 589b40810a ci : dummy commit to trigger CI Georgi Gerganov 2025-02-03 16:32:48 +02:00
  • 7ffcd05267 ruby : Make context accept initial parameters, API to retrieve a segment and more (#2749) KITAITI Makoto 2025-01-21 16:39:54 +09:00
  • 7a423f1c00 whisper.objc : fix build and CI Corey Earwood 2025-01-18 03:06:06 -07:00
  • 99b011a9f5 talk-llama : sync llama.cpp Georgi Gerganov 2025-01-14 09:53:50 +02:00
  • 19d95f9f9a sync : ggml Georgi Gerganov 2025-01-14 09:50:06 +02:00
  • d5ef1737d8 GGUF: C++ refactor, backend support, misc fixes (skip) (llama/11030) Johannes Gäßler 2025-01-14 09:31:07 +02:00
  • 1deb41f0e7 ggml : add opencl backend (skip) (llama/10693) lhez 2025-01-14 09:24:03 +02:00
  • 2425caf4fd cuda : CUDA Graph Compute Function Refactor (precursor for performance improvements) (llama/11042) Andreas Kieslinger 2025-01-13 16:45:53 +01:00
  • a4b00bcaaf ggml : do not define GGML_USE_CUDA when building with GGML_BACKEND_DL (llama/11211) Radoslav Gerganov 2025-01-13 13:31:41 +02:00
  • cdb8aa2f2e Vulkan: Fix float16 use on devices without float16 support + fix subgroup_size_control validation error (llama/11161) 0cc4m 2025-01-10 06:39:33 +01:00
  • 06209f6683 llama: add support for QRWKV6 model architecture (llama/11001) Molly Sophia 2025-01-10 09:58:08 +08:00
  • c3235bd81e SYCL: Refactor ggml_sycl_compute_forward (llama/11121) Akarshan Biswas 2025-01-10 05:43:03 +05:30
  • 262d0abc87 fix: add missing msg in static_assert (llama/11143) hydai 2025-01-09 04:03:28 +08:00
  • 124eec1664 llamafile : ppc64le MMA INT8 implementation (llama/10912) amritahs-ibm 2025-01-08 16:24:19 +05:30
  • b08c3a88c8 Disable GL_KHR_cooperative_matrix Vulkan extension if not available. (llama/11117) Mathieu Baudier 2025-01-08 09:18:13 +01:00
  • 0afce25a69 fix: Vulkan shader gen binary path when Cross-compiling (llama/11096) ag2s20150909 2025-01-08 16:17:29 +08:00
  • acdbe58631 GGUF: C++ refactor, backend support, misc fixes (llama/11030) Johannes Gäßler 2025-01-07 18:01:58 +01:00
  • 09fabffdf5 ggml-backend : only offload from host buffers (fix) (llama/11124) Diego Devesa 2025-01-07 16:11:57 +01:00
  • 3988d6396b ggml-backend : only offload from host buffers (llama/11120) Diego Devesa 2025-01-07 12:38:05 +01:00
  • c8c63eeec0 rpc : code cleanup (llama/11107) Radoslav Gerganov 2025-01-07 08:37:02 +02:00
  • abf7f24410 SYCL: Use get_multi_ptr instead of deprecated get_pointer in wkv6 (llama/11087) Akarshan Biswas 2025-01-07 11:56:07 +05:30
  • 341f5c28e6 CUDA: add BF16 support (llama/11093) Johannes Gäßler 2025-01-06 02:33:52 +01:00
  • 5377099524 Vulkan: Add device-specific blacklist for coopmat for the AMD proprietary driver (llama/11074) 0cc4m 2025-01-04 21:09:59 +01:00
  • dcbb375779 Support for models with non-512-aligned tensors over RPC. (llama/11047) matt23654 2025-01-04 16:10:30 +00:00
  • 4334c71aed fix: Vulkan shader gen binary path (llama/11037) Gilad S. 2025-01-04 10:17:31 +02:00
  • e875a82473 ggml : allow loading backend with env variable (ggml/1059) Radoslav Gerganov 2025-01-05 09:50:37 +02:00
  • 507e230f1e scripts : sync opencl, gguf Georgi Gerganov 2025-01-14 09:42:16 +02:00
  • eb68324c86 whisper : fix gpu device selection (#2728) Georgi Gerganov 2025-01-13 13:11:37 +02:00
  • e940fbf283 server : fix build (#2718) Georgi Gerganov 2025-01-13 08:57:33 +02:00
  • 35d0e02c72 talk-llama : sync llama.cpp (#2709) Georgi Gerganov 2025-01-13 08:55:48 +02:00
  • 45d3faf961 server : generate unique tmp filenames (#2718) NETZkultur GmbH 2025-01-13 07:55:21 +01:00
  • 2ab2eb5110 whisper : add whisper_full_get_segment_no_speech_prob_from_state (#2716) Sandro Hanea 2025-01-09 15:21:07 +01:00
  • b82d305282 readme : add docker instructions (#2711) Jayant 2025-01-07 12:20:51 +01:00
  • 885e31368d docs: Fix main -> whisper-cli in download scripts (#2707) Adam Jones 2025-01-06 13:17:57 +00:00
  • 8a9ad7844d release : v1.7.4 v1.7.4 Georgi Gerganov 2025-01-06 15:13:48 +02:00
  • eb874b3a3c ci : cont Georgi Gerganov 2025-01-06 10:46:10 +02:00
  • eb78e3a3f1 ci : fix ubuntu runner names Georgi Gerganov 2025-01-06 09:29:10 +02:00
  • ece3ff88f6 cli : fix segfault on missing argument (#2700) Yusuf Redžić 2025-01-04 09:47:41 +01:00
  • 9366544991 ci : fix arm builds Georgi Gerganov 2025-01-03 16:24:02 +02:00
  • 95583942ed sync : ggml Georgi Gerganov 2025-01-03 14:11:23 +02:00
  • 2e93cb6a2f ggml : do not install metal source when embed library (ggml/1054) Georgi Gerganov 2025-01-03 14:11:20 +02:00
  • de5cd60d1c metal : avoid uint (llama/11019) Georgi Gerganov 2025-01-03 11:26:14 +02:00
  • 3fcba3e58b ggml : fixes for AVXVNNI instruction set with MSVC and Clang (llama/11027) Srihari-mcw 2024-12-31 19:53:33 +05:30
  • cea5f1c52f vulkan: optimize mul_mat for small values of N (llama/10991) Jeff Bolz 2024-12-30 11:27:11 -06:00
  • 2112462db4 vulkan: im2col and matmul optimizations for stable diffusion (llama/10942) Jeff Bolz 2024-12-29 03:16:34 -06:00
  • fc84ecd445 vulkan: Use push constant offset to handle misaligned descriptors (llama/10987) Jeff Bolz 2024-12-29 02:35:11 -06:00
  • 8de1e99907 vulkan: multi-row k quants (llama/10846) Eve 2024-12-26 10:54:44 -05:00
  • 499af9294a examples, ggml : fix GCC compiler warnings (llama/10983) Peter 2024-12-27 00:59:11 +11:00
  • bcf937c216 ggml : more perfo with llamafile tinyblas on x86_64 (llama/10714) Djip007 2024-12-24 18:54:49 +01:00
  • b8d90953d7 ggml : use wstring for backend search paths (llama/10960) Diego Devesa 2024-12-24 04:05:27 +01:00
  • 60a422147b ggml : fix arm enabled features check (llama/10961) Diego Devesa 2024-12-24 04:05:17 +01:00
  • 3387415bad ggml : fix const usage in SSE path (llama/10962) Diego Devesa 2024-12-23 20:25:52 +01:00
  • 536ca3ec89 ggml : fix run-time on FreeBSD in get_executable_path() (llama/10948) yuri@FreeBSD 2024-12-22 16:20:11 -08:00
  • a4bb983190 vulkan: build fixes for 32b (llama/10927) Jeff Bolz 2024-12-22 03:44:01 -06:00
  • 39c205f555 vulkan: optimize coopmat2 dequant functions (llama/10855) Jeff Bolz 2024-12-21 01:04:45 -06:00
  • 6d502f33dc ggml-cpu: replace NEON asm with intrinsics in ggml_gemv_q4_0_4x8_q8_0() (llama/10874) Adrien Gallouët 2024-12-21 00:33:37 +01:00
  • 5ea27d089d SYCL: Migrate away from deprecated ggml_tensor->backend (llama/10840) Akarshan Biswas 2024-12-20 21:01:28 +05:30
  • 1462d92588 ggml : add test for SVE and disable when it fails (llama/10906) Diego Devesa 2024-12-20 13:31:28 +01:00
  • 7ba1a41f47 ggml: fix arm build with gcc (llama/10895) Adrien Gallouët 2024-12-19 14:20:41 +01:00
  • 5ea088636f ggml : fix arm build (llama/10890) Diego Devesa 2024-12-18 23:21:42 +01:00
  • f32ddb3b1c tts : add OuteTTS support (llama/10784) Georgi Gerganov 2024-12-18 19:27:21 +02:00
  • 79b75ece03 tests: add tests for GGUF (llama/10830) Johannes Gäßler 2024-12-17 19:09:35 +01:00