Commit Graph

Select branches

Hide Pull Requests

fable5/prefetch-experts

fable5/vram-ondemand

fable5/vram-ondemand-rebase-20260909

fable5/vram-ondemand-rebased-20260726

gemini

Mono Color

• c5d70f5c9e ggml : optimize rope function to avoid call powf in the tight loop (#807) Howard Su 2023-04-14 14:24:52 +08:00
• be87b6ed20 perplexity : add support for batch size to --perplexity (#407) Gary Linscott 2023-04-13 14:50:42 -07:00
• 0e07e6a839 common : remove unnecessary includes (#947) CRD716 2023-04-13 10:39:25 -05:00
• a3a2a0eda8 ggml : add GGML_DEFAULT_N_THREADS Georgi Gerganov 2023-04-13 18:36:40 +03:00
• d990e3fffc ggml : speed-up ggml_vec_dot_q4_1() ARM_NEON + 32-bit ARM support (#900) Georgi Gerganov 2023-04-13 18:32:36 +03:00
• 9190e8eac8 llama : merge llama_internal.h into llama.h Georgi Gerganov 2023-04-13 18:04:45 +03:00
• c85980acd0 gitignore : benchmark Georgi Gerganov 2023-04-13 18:01:22 +03:00
• 6232f2d7fd ggml : optimize non-SIMD Q4_0 vector dot product (#703) Stephan Walter 2023-04-13 14:59:50 +00:00
• 6c248707f5 ggml : introduce GGML_ALIGNED_MALLOC/GGML_ALIGNED_FREE macros (#884) Pavol Rusnak 2023-04-13 16:08:32 +02:00
• 8cda5c981d fix whitespace (#944) CRD716 2023-04-13 09:03:57 -05:00
• ec29272175 readme : remove python 3.10 warning (#929) CRD716 2023-04-13 08:59:53 -05:00
• 7e941b95eb readme : llama node binding (#911) Genkagaku.GPT 2023-04-13 21:54:27 +08:00
• c729ff730a flake.nix: add all binaries from bin (#848) Pavol Rusnak 2023-04-13 15:49:05 +02:00
• 4579af95e8 zig : update build.zig (#872) Juddandzjli2019 2023-04-13 21:43:22 +08:00
• 8c3ffc2f04 ggml : update cblas_sgemm columns var to be more reasonable (#838) Vladimir 2023-04-13 15:24:30 +02:00
• 107980d970 examples : add -n to alpaca and gpt4all scripts (#706) niansa/tuxifan 2023-04-13 15:03:39 +02:00
• 585d91a156 cmake : add explicit F16C option (x86) (#576) anzz1 2023-04-13 15:48:21 +03:00
• 95ea26f6e9 benchmark : add tool for timing q4_0 matrix multiplication (#653) SebastianApelandGeorgi Gerganov 2023-04-13 14:46:23 +02:00
• 82d146df9b do not force the prompt file to end with a new line (#908) Pavol Rusnak 2023-04-13 11:33:16 +02:00
• e7f6997f89 Don't crash on ftype (formerly f16) == 4 (#917) Stephan Walter 2023-04-12 15:06:16 +00:00
• f76cb3a34d readme : change "GPU support" link to discussion Georgi Gerganov 2023-04-12 14:48:57 +03:00
• 782438070f readme : update hot topics with link to "GPU support" issue Georgi Gerganov 2023-04-12 14:31:12 +03:00
• 4dbbd40750 readme: link to sha256sums file (#902) Nicolai Weitkemper 2023-04-12 08:46:20 +02:00
• 8b679987cd Fix whitespace, add .editorconfig, add GitHub workflow (#883) Pavol Rusnak 2023-04-11 21:45:44 +02:00
• 3e6e70d8e8 Add enum llama_ftype, sync ggml_type to model files (#709) Stephan Walter 2023-04-11 15:03:51 +00:00
• 2663d2c678 Windows fixes (#890) comex 2023-04-11 06:19:54 -07:00
• a0caa34b16 Add BAIR's Koala to supported models (#877) qouoq 2023-04-11 04:41:53 +08:00
• 461ba9e66e ggml : fix WASM build Georgi Gerganov 2023-04-10 23:20:01 +03:00
• c3ac702e5e ggml : add ggml_cont() + optimize ggml_cpy() for contiguous dst Georgi Gerganov 2023-04-10 22:40:28 +03:00
• 9d634ef452 ggml : remove trailing whitespaces Georgi Gerganov 2023-04-10 19:32:45 +03:00
• d9a239c410 Simplify to include lower-case windows.h always, fix compile on mingw32 (#747) Marco Matthies 2023-04-10 19:57:59 +02:00
• 684da25926 ggml : fix quantize_row_q4_1() ARM_NEON (close #876) Georgi Gerganov 2023-04-10 19:29:48 +03:00
• 180b693a47 Print model version. comex 2023-04-08 13:08:21 -07:00
• f963b63afa Rewrite loading code to try to satisfy everyone: comexandPavol Rusnak 2023-04-08 12:24:37 -07:00
• aaf3b23deb fix for windows utf-8 input (#840) Tomáš Pazdiora 2023-04-08 17:49:39 +02:00
• f2d1c47294 cmake should link openblas properly with -lopenblas like how it's done in the makefile (#839) eiery 2023-04-08 07:15:17 -04:00
• 317fb12fbd Add new binaries to flake.nix (#847) lon 2023-04-08 07:04:23 -03:00
• 62cfc54f77 Add quantize-stats command for testing quantization (#728) unbounded 2023-04-08 00:09:18 +02:00
• 698f7b5d63 make : add libllama.so target for llama-cpp-python (#797) bhubbb 2023-04-08 02:11:58 +10:00
• c1950c3431 zig : don't link examples/common.cpp for non-example (#814) iacore 2023-04-07 16:05:29 +00:00
• 4953e9007f llama : always sort logits before nucleus sampling (#812) Ivan Stepanov 2023-04-07 19:02:12 +03:00
• cc9cee8e9e Do not crash when it has nothing to say. (#796) Sergey Alirzaev 2023-04-06 17:59:11 +02:00
• d2beca95dc Make docker instructions more explicit (#785) Pavol Rusnak 2023-04-06 08:56:58 +02:00
• eeaa7b0492 ggml : multi-thread ggml_rope() (~3-4 times faster on M1) (#781) Georgi Gerganov 2023-04-05 22:11:03 +03:00
• 986b6ce9f9 ggml, llama : avoid heavy V transpose + improvements (#775) Georgi Gerganov 2023-04-05 22:07:33 +03:00
• 3416298929 Update README.md Georgi Gerganov 2023-04-05 19:54:30 +03:00
• 5a8c4f6240 llama : define non-positive top_k; top_k range check (#779) Ivan StepanovandGeorgi Gerganov 2023-04-05 19:20:05 +03:00
• ff05d05c96 miku.sh : add executable bit (#780) at8u 2023-04-05 15:59:13 +00:00
• 62b3e81aae media : add logos and banners Georgi Gerganov 2023-04-05 18:58:06 +03:00
• 8d10406d6e readme : change logo + add bindings + add uis + add wiki Georgi Gerganov 2023-04-05 18:56:20 +03:00
• ed1c214e66 zig : add build.zig (#773) iacoreandLocria Cyber 2023-04-05 15:06:02 +00:00
• 0c44427df1 make : missing host optimizations in CXXFLAGS (#763) Ivan Stepanov 2023-04-05 17:38:37 +03:00
• 594cc95fab readme : update with CMake and windows example (#748) Adithya Balaji 2023-04-05 16:36:12 +02:00
• 88ed5761b8 examples : add Miku.sh (#724) at8u 2023-04-05 14:32:42 +00:00
• 58c438cf7d Add Accelerate/BLAS when using Swift (#765) Andrew Duffy 2023-04-05 11:44:24 +01:00
• 53dbba7695 Windows: reactive sigint handler after each Ctrl-C (#736) mgroeber9110 2023-04-03 18:00:55 +02:00
• 437e77855a 10+% performance improvement of ggml_vec_dot_q4_0 on AVX2 (#654) SebastianApel 2023-04-03 09:52:28 +02:00
• cd7fa95690 Define non-positive temperature behavior (#720) Ivan Stepanov 2023-04-03 03:19:04 +03:00
• a0c0516416 Remove torch GPU dependencies from the Docker.full image (#665) bsilvereagle 2023-04-02 15:13:03 -07:00
• d8d4e865cd Add a missing step to the gpt4all instructions (#690) Thatcher Chamberlin 2023-04-02 06:48:57 -04:00
• e986f94829 Added api for getting/setting the kv_cache (#685) Christian FalchandPavol Rusnak 2023-04-02 12:23:04 +02:00
• c0bb1d3ce2 ggml : change ne to int64_t (#626) Marian Cepok 2023-04-02 12:21:31 +02:00
• 6e7801d08d examples : add gpt4all script (#658) Leonardo Neumann 2023-04-02 04:56:20 -03:00
• 81040f10aa llama : do not allocate KV cache for "vocab_only == true" (#682) Stephan Walter 2023-04-02 07:18:53 +00:00
• c4f89d8d73 make : use -march=native -mtune=native on x86 (#609) Fabian 2023-04-02 09:17:05 +02:00
• 5b70e7de4c fix default params for examples/main (#697) Murilo Santana 2023-04-01 23:41:12 -03:00
• a717cba844 py: huggingface -> Hugging Face (#686) Ikko Eltociear Ashimine 2023-04-02 01:38:18 +09:00
• d0a7f742e7 readme: replace termux links with homepage, play store is deprecated (#680) rimoliga 2023-04-01 11:57:30 -03:00
• 0d054e292e Show error message when -f fails Slaren 2023-03-31 20:03:48 +02:00
• 3525899277 Enable -std= for cmake builds, fix warnings (#598) Stephan Walter 2023-03-31 19:19:16 +00:00
• 1d08882afa Optimize AVX2 ggml_vec_dot_q4_0 (#642) slaren 2023-03-31 17:55:52 +02:00
• 02c5b27e91 Add AVX acceleration (#617) perserk 2023-03-31 16:55:44 +05:00
• cbef542879 py : cleanup the code Pavol Rusnak 2023-03-29 21:31:24 +02:00
• 9733104be5 drop quantize.py (now that models are using a single file) Pavol Rusnak 2023-03-31 00:52:06 +02:00
• 3df890aef4 readme : update supported models Georgi Gerganov 2023-03-30 22:31:54 +03:00
• ee0c40dd6d Introduce GGML migration tool for new file format Justine Tunney 2023-03-30 05:42:56 -07:00
• 6f23ba5ee2 Ensure --mlock works properly with mmap() support Justine Tunney 2023-03-30 01:53:36 -07:00
• 78ca9838ee Make loading weights 10-100x faster Justine Tunney 2023-03-29 13:51:37 -07:00
• a017390358 Initial windows support (untested) Slaren 2023-03-29 22:22:36 +02:00
• ac184d5147 Always initialize mm_addr and mm_length in llama_model Slaren 2023-03-29 08:53:14 +02:00
• 276e5b7811 Unmap the file in llama_free Slaren 2023-03-29 08:31:26 +02:00
• d68c5dc435 Make mmap_file static Slaren 2023-03-29 06:18:18 +02:00
• 64bde3ffd4 Fix ggml_init_params in quantize Slaren 2023-03-29 05:38:57 +02:00
• c03ae8dca1 Add mmap support for model files Slaren 2023-03-29 02:03:43 +02:00
• 3bcc129ba8 cmake : properly invoke CTest (#629) Stephan Walter 2023-03-30 17:56:59 +00:00
• a4755cf288 Remove unused variable (#607) Casey Primozic 2023-03-30 10:53:35 -07:00
• 1f0414feec make : fix darwin f16c flags check (#615) david raistrick 2023-03-30 13:34:45 -04:00
• 77efdf5a50 ggml : fix NEON signs (close #620, #622) Georgi Gerganov 2023-03-30 20:27:32 +03:00
• ed3c680bcd Fix GGML_F32Cx8_STORE in AVX without F16C path (#619) slaren 2023-03-30 11:16:30 +02:00
• 9cbc404ba6 ci : re-enable AVX512 testing (Windows-MSVC) (#584) anzz1 2023-03-29 23:44:39 +03:00
• b51c717d5c ggml : init time on first ggml_init() call Georgi Gerganov 2023-03-29 22:15:34 +03:00
• 0ba76c1e73 llama : fix compile warnings when reading the vocab Georgi Gerganov 2023-03-29 22:13:12 +03:00
• cea1c85948 ggml : add ARM_NEON dequantize_row_q4_1() Georgi Gerganov 2023-03-29 22:10:01 +03:00
• f202ada131 ggml : add ARM_NEON quantize_row_q4_1() Georgi Gerganov 2023-03-29 22:03:02 +03:00
• 3b44d30d9b ggml : add ARM_NEON ggml_vec_dot_q4_1() Georgi Gerganov 2023-03-29 21:47:33 +03:00
• 61cbfff5c9 rename convert_ggml_to_pth.py -> convert-ggml-to-pth.py (#600) Pavol Rusnak 2023-03-29 20:09:25 +02:00
• d9ad104440 Create chat-13B.bat (#592) Thérenceandanzz1 2023-03-29 19:21:09 +02:00
• b467702b87 readme : fix typos Georgi Gerganov 2023-03-29 19:38:31 +03:00
• 516d88e75c readme : add GPT4All instructions (close #588) Georgi Gerganov 2023-03-29 19:37:20 +03:00
• 53635c081c py : add GPT4All conversion script Georgi Gerganov 2023-03-29 19:29:26 +03:00