Commit Graph

  • c96b0a938e ggml : group all experts in a single ggml_mul_mat_id (llama/6505) slaren 2024-04-18 15:18:48 +02:00
  • c97796aa0f ggml : fix llamafile sgemm wdata offsets (llama/6710) Georgi Gerganov 2024-04-16 23:50:22 +03:00
  • 7a4f7d825e ggml : add llamafile sgemm (llama/6414) Justine Tunney 2024-04-16 14:55:30 -04:00
  • fdb2c87350 llama : add qwen2moe (llama/6074) Shijie 2024-04-16 23:40:48 +08:00
  • 98c0b77e0c fix mul_mat_id() for new input, make the ut pass (llama/6682) Neo Zhang Jianyu 2024-04-15 17:12:26 +08:00
  • 9d6d50d933 Added support for GGML_OP_CLAMP in Metal (llama/6662) Dave 2024-04-14 07:14:19 -04:00
  • c1320c1f0c fix memcpy() crash, add missed cmd in guide, fix softmax (llama/6622) Neo Zhang Jianyu 2024-04-14 10:42:29 +08:00
  • 66aaf03a7a CUDA: fix matrix multiplication logic for tests (llama/6667) Johannes Gäßler 2024-04-14 00:21:55 +02:00
  • 00a0947c65 metal : unify mul_mv_id kernels (llama/6556) slaren 2024-04-12 18:13:20 +02:00
  • 60f3713026 llama : add gguf_remove_key + remove split meta during quantize (llama/6591) jiez 2024-04-12 18:45:06 +08:00
  • 37e6757453 feat: implemented sigmoid function (ggml/806) Justina Cho 2024-05-01 14:44:26 -07:00
  • 8dcefdf4a9 build: fix and ignore msvc warnings (ggml/805) Borislav Stanimirov 2024-04-25 17:24:07 +03:00
  • 73d13ad19a ggml : expose SSE3 and SSSE3 for MSVC when AVX is available (#2128) Przemysław Pawełczyk 2024-05-08 17:33:43 +02:00
  • b6680fab50 build : improve disabling AVX-512 (#2129) Przemysław Pawełczyk 2024-05-08 17:32:43 +02:00
  • f760756078 minor: add CMakeSettings.json to gitignore (#2094) Borislav Stanimirov 2024-05-08 11:03:21 +03:00
  • 58210d6a76 examples : fix node compilation (#2115) Pedro Probst 2024-05-02 18:52:55 -03:00
  • 8fac6455ff make : change GNU make default CXX from g++ to c++ (#2100) Przemysław Pawełczyk 2024-04-28 23:54:21 +02:00
  • 22b6598cc9 Remove unnecessary memory reallocation in fft (#2080) goldwaving 2024-04-28 15:06:12 -02:30
  • 858452d58d models : disable old script (#2079) Georgi Gerganov 2024-04-24 14:56:30 +03:00
  • 7f85e1d7fd whisper : more prominent log message for sub-1s audio (#2065) Georgi Gerganov 2024-04-24 14:45:27 +03:00
  • b0c3cbf2e8 main : pass nullptr when regex is empty (#2070) Georgi Gerganov 2024-04-17 12:23:47 +03:00
  • a750868428 readme : add up-to-date repository for Python bindings (#2063) AIWintermuteAI 2024-04-16 19:15:52 +08:00
  • 7395c70a74 release : v1.5.5 v1.5.5 Georgi Gerganov 2024-04-16 14:08:31 +03:00
  • 9fab28135c server : add dtw (#2044) Emmanuel Schmidbauer 2024-04-15 15:16:58 -04:00
  • 08d3eef97d build : fix embedded Metal library generation (#2045) Didzis Gosko 2024-04-15 20:23:05 +03:00
  • 1b5439a6c2 node : support no timestamps (#2048) Pedro Probst 2024-04-15 14:03:34 -03:00
  • c7f95b7ca2 build : detect AVX512 in Makefile, add AVX512 option in CMake (#2043) Didzis Gosko 2024-04-15 20:02:09 +03:00
  • 5c554c04ff whisper.nvim : fix missing reference to "model" variable (#2049) Kendrick Taylor 2024-04-15 09:41:28 -07:00
  • c383f091a1 whisper : update grammar-parser.cpp (#2058) Ikko Eltociear Ashimine 2024-04-16 01:40:27 +09:00
  • 8f253ef3af sync : ggml Georgi Gerganov 2024-04-09 20:27:55 +03:00
  • c7dc37f97c license : update copyright notice + add AUTHORS Georgi Gerganov 2024-04-09 20:27:44 +03:00
  • 526332873b llama : add Command R Plus support (llama/6491) Carolinabanana 2024-04-09 09:16:13 +01:00
  • 1d2721ca72 remove row=1 cond (llama/6532) Abhilash Majumder 2024-04-08 13:56:01 +05:30
  • 219e601dab support/fix OPs GGML_TYPE_IQ4_NL, GGML_TYPE_IQ4_XS, GGML_TYPE_IQ3_XXS, GGML_TYPE_IQ3_S, GGML_TYPE_IQ2_XXS, GGML_TYPE_IQ2_XS, GGML_TYPE_IQ2_S, GGML_TYPE_IQ1_S, GGML_TYPE_IQ1_M (llama/6521) Neo Zhang Jianyu 2024-04-07 10:55:59 +08:00
  • 3b8aade3c2 scripts : update sync Georgi Gerganov 2024-04-09 20:25:50 +03:00
  • 52ccd4a3a8 files : rename ./extra to ./scripts Georgi Gerganov 2024-04-09 20:12:17 +03:00
  • 5275074d37 whisper : fix DTW memory access (#2012) Brad Murray 2024-04-09 11:38:19 -04:00
  • c15b4cda7d common : fix file-handle leak in read_wav() (#2026) ulatekh 2024-04-09 08:34:34 -07:00
  • d3cfb6ca2b main : set stdin to binary mode on Windows (#2025) Rotem Dan 2024-04-09 18:33:32 +03:00
  • 956ef860bc cmake : support for CPU BLAS build via Intel MKL (#2024) slashlib 2024-04-09 18:32:46 +03:00
  • 671b4bde6c main : allow a response-file as the sole parameter (#2019) ulatekh 2024-04-09 08:31:16 -07:00
  • c8eeb93a6a whisper : suppress tokens with a regex (#1997) ulatekh 2024-04-09 08:27:28 -07:00
  • 319fe5146e cmake : create solution folders (#2004) ulatekh 2024-04-09 08:23:33 -07:00
  • 13c22321d1 sync : ggml Georgi Gerganov 2024-04-07 17:04:56 +03:00
  • ccbe9d5676 extra : sync grammar-parser Georgi Gerganov 2024-04-07 17:04:22 +03:00
  • 81a3c41aa0 talk-llama : sync llama.cpp Georgi Gerganov 2024-04-07 16:21:08 +03:00
  • a50207c65d sync : ggml Georgi Gerganov 2024-04-07 16:18:11 +03:00
  • 97878e53fd sync : llama.cpp (skip) Georgi Gerganov 2024-04-06 17:50:21 +03:00
  • 61b05815e0 Fixed minor bug when enabling FP16 for non intel targets (llama/6464) Ouadie EL FAROUKI 2024-04-05 14:35:06 +01:00
  • 1dce94cf26 ggml : mul_mat_id use the same tensor for all the experts (llama/6387) slaren 2024-04-03 15:07:05 +02:00
  • f12e982c0b Disable iqx on windows as WA (llama/6435) Meng, Hengyu 2024-04-03 10:34:40 +08:00
  • fa966b9b40 Vulkan k-quant mmq and ggml-backend offload functionality (llama/6155) 0cc4m 2024-03-29 17:29:21 +01:00
  • b83a9fc9d3 fix set main gpu crash (llama/6339) Neo Zhang Jianyu 2024-03-28 08:55:24 +08:00
  • 3adbf2fb03 ggml : fix bounds checking of zero size views (llama/6347) slaren 2024-03-27 15:07:50 +01:00
  • 700d146127 backend : fix typo in scheduler documentation (ggml/781) Daniel Bevenius 2024-04-03 22:57:20 +02:00
  • a74fde9b4c extra : sync ggml-cuda folder Georgi Gerganov 2024-04-07 16:10:44 +03:00
  • 1d7657f409 ggml: bypass code incompatible with CUDA < 11.1 (#2020) Slava Primenko 2024-04-04 14:49:24 +02:00
  • ac283dbce7 ci : add building in MSYS2 environments (Windows) (#1994) Przemysław Pawełczyk 2024-03-30 08:20:20 +01:00
  • 1e8f28c42a build : use pkg-config for OpenBLAS (#1778) Przemysław Pawełczyk 2024-03-29 14:53:26 +01:00
  • fc366b807a main : add command-style grammar (#1998) ulatekh 2024-03-28 03:02:10 -07:00
  • 9fb308d90f make : add grammar parser to common objects Georgi Gerganov 2024-03-28 11:59:48 +02:00
  • 2948c740a2 sync : ggml (#2001) Georgi Gerganov 2024-03-27 18:55:10 +02:00
  • 1558ec5a16 whisper : improve handling of prompts (#1981) Georgi Gerganov 2024-03-25 14:48:19 +02:00
  • ba69578828 whisper : add whisper_token_count helper gg/prompt-tokens Georgi Gerganov 2024-03-25 14:46:07 +02:00
  • fff24a0148 whisper : improve support for distil-large-v3 (#1982) Sanchit Gandhi 2024-03-21 22:23:30 +05:30
  • 5c2c07d479 whisper : improve handling of prompts Georgi Gerganov 2024-03-21 08:00:11 +02:00
  • 48a145207e ruby : fix build (#1980) Georgi Gerganov 2024-03-21 07:40:09 +02:00
  • 79d5765e7e docker : libcuda.so.1 in PATH (#1966) Tiago Fassoni 2024-03-20 13:45:15 -03:00
  • 04e48094e4 readme : add Fedora dependencies (#1970) Mohammadreza Hendiani 2024-03-20 20:12:11 +03:30
  • 741abb162c whisper : token-level timestamps with DTW (#1485) denersc 2024-03-20 13:25:26 -03:00
  • e7794a868f examples : rename --audio-context to --audio-ctx per help text (#1953) Jo Liss 2024-03-18 15:53:33 +00:00
  • 725350d4ea whisper : set outputs from conv graph (#1959) Georgi Gerganov 2024-03-16 17:30:55 +02:00
  • 906c73b219 alloc : fix allocation data of pre-allocated leafs slaren 2024-03-16 15:47:14 +01:00
  • 00d80ff965 cmake : copy ggml-common.h to bin Georgi Gerganov 2024-03-16 17:15:21 +02:00
  • 66df44b0b7 alloc : fix allocation data of pre-allocated leafs gg/alloc-enc-results slaren 2024-03-16 15:47:14 +01:00
  • 1b553b9817 gitignore : .vimspector.json Georgi Gerganov 2024-03-16 16:26:35 +02:00
  • c6c94de43a whisper : allocate encoder results in dedicated buffer Georgi Gerganov 2024-03-16 16:02:48 +02:00
  • de4d067f1e talk-llama : sync llama.cpp Georgi Gerganov 2024-03-15 14:21:59 +02:00
  • e715f6a601 sync : ggml Georgi Gerganov 2024-03-15 14:12:19 +02:00
  • f60ccfd83b update examples and tests slaren 2024-03-14 16:45:27 +01:00
  • 3753a2b2a8 ggml : add ggml-common.h Georgi Gerganov 2024-03-14 17:16:45 +02:00
  • 592dd25615 ggml : designate enum vals for integer types (llama/6050) Georgi Gerganov 2024-03-14 12:38:37 +02:00
  • c8709d4604 metal : build metallib + fix embed path (llama/6015) Georgi Gerganov 2024-03-14 11:55:23 +02:00
  • 8932c2d6ce llama : add pipeline parallelism support (llama/6017) slaren 2024-03-13 18:54:21 +01:00
  • 2bddfdd7c8 Update get version (llama/6025) AidanBeltonS 2024-03-13 13:17:54 +00:00
  • 46e3c3f112 ggml : reuse quantum structs across backends (llama/5943) Georgi Gerganov 2024-03-12 14:27:20 +02:00
  • ef24ae0c7d ggml : fix UB in IQ2_S and IQ3_S (llama/6012) Georgi Gerganov 2024-03-12 13:49:55 +02:00
  • a753926f02 sycl : update IQ1_S kernels (WIP - not working!) (llama/5995) Georgi Gerganov 2024-03-12 11:15:05 +02:00
  • 9dc60fc02d 1.5 bit: we can do even better (llama/5999) Kawrakow 2024-03-11 16:53:15 +01:00
  • d73a63629e ggml, ci : Windows ARM runner and build fixes (llama/5979) Michael Podvitskiy 2024-03-11 10:28:51 +01:00
  • f79d0d4f74 Better 1.5 bit quantization (llama/5971) Kawrakow 2024-03-11 07:51:49 +01:00
  • 4f88940ff6 Add q3_s and q1_s (llama/5886) Abhilash Majumder 2024-03-11 10:27:56 +05:30
  • 7bdb1de9ec metal : move mm_id indices to shared mem (llama/5982) Georgi Gerganov 2024-03-10 23:12:48 +02:00
  • 653d2e8ff9 ggml : fix unnecessary f32 -> f16 -> f32 casts (mmla) (llama/5951) Georgi Gerganov 2024-03-09 17:36:20 +02:00
  • 2fef660d0a ggml : remove old quantization functions (llama/5942) Georgi Gerganov 2024-03-09 15:53:59 +02:00
  • 24eba5a2ff ggml : add ggml-common.h to deduplicate shared code (llama/5940) Georgi Gerganov 2024-03-09 12:47:57 +02:00
  • 6e9d3aa32d llama : support Mamba Selective State Space Models (llama/5328) compilade 2024-03-08 17:31:00 -05:00
  • 9ae0d18856 extra : update sync scripts after ggml-common.h Georgi Gerganov 2024-03-15 14:00:53 +02:00
  • a56f435fd4 whisper : document whisper_batch.n_seq_id (#1942) Josh Bleecher Snyder 2024-03-10 07:55:22 -07:00
  • ec166499d8 whisper : improve beam search candidate diversity (#1947) Josh Bleecher Snyder 2024-03-10 07:54:43 -07:00