]> git.djapps.eu Git - pkg/ggml/sources/llama.cpp/shortlog
pkg/ggml/sources/llama.cpp
2023-11-05 KerfuffleAllow common process_escapes to handle \x sequences...
2023-11-05 Thái Hoàng Tâmserver : fix typo for --alias shortcut from -m to ...
2023-11-05 Jared Van Bortelcuda : fix disabling device with --tensor-split 1,0...
2023-11-05 Meng Zhangllama : mark LLM_ARCH_STARCODER as full offload support...
2023-11-05 Evecmake : MSVC instruction detection (fixed up #809)...
2023-11-05 Eveci : use intel sde when ci cpu doesn't support avx512...
2023-11-05 slarencuda : revert CUDA pool stuff (#3944)
2023-11-04 Kerfufflegguf-py: Support 01.AI Yi models (#3943)
2023-11-03 Peter Sugiharametal : round up to 16 to fix MTLDebugComputeCommandEnc...
2023-11-03 Xiao-Yong Jinggml-metal: fix yarn rope (#3937)
2023-11-03 slarenggml-cuda : move row numbers to x grid dim in mmv kerne...
2023-11-03 Georgi Gerganovspeculative : change default p_accept to 0.5 + CLI...
2023-11-03 Georgi Gerganovcommon : YAYF (yet another YARN fix) (#3925)
2023-11-03 cebtenzzrellama : change yarn_ext_factor placeholder to -1 (...
2023-11-02 Kerfufflecuda : add ROCM aliases for CUDA pool stuff (#3918)
2023-11-02 Andreicmake : fix relative path to git submodule index (...
2023-11-02 Georgi Gerganovreadme : add notice about #3912
2023-11-02 Georgi Gerganovcuda : fix const ptrs warning causing ROCm build issues...
2023-11-02 Oleksii Maryshchenkocuda : use CUDA memory pool with async memory allocatio...
2023-11-02 Georgi Gerganovgguf : print error for GGUFv1 files (#3908)
2023-11-02 slarencmake : disable LLAMA_NATIVE by default (#3906)
2023-11-02 Georgi Gerganovgguf : remove special-case code for GGUFv1 (#3901)
2023-11-02 Georgi Gerganovllm : prevent from 1-D tensors being GPU split (#3697)
2023-11-02 cebtenzzrebuild : link against build info instead of compiling...
2023-11-02 Georgi Gerganovcuda : check if this fixes Pascal card regression ...
2023-11-02 Georgi Gerganovmetal : fix build errors and kernel sig after #2268...
2023-11-02 cebtenzzrecuda : fix RoPE after #2268 (#3897)
2023-11-01 cebtenzzrellama : fix llama_context_default_params after #2268...
2023-11-01 slarenggml-cuda : compute ptrs for cublasGemmBatchedEx in...
2023-11-01 cebtenzzrellama : implement YaRN RoPE scaling (#2268)
2023-11-01 Georgi Gerganovllm : fix llm_build_kqv taking unused tensor (benign...
2023-11-01 Georgi Gerganovllm : fix falcon norm after refactoring (#3837)
2023-11-01 Georgi Gerganovmetal : multi-simd softmax (#3710)
2023-11-01 Georgi Gerganovcommon : minor (#3715)
2023-11-01 Georgi Gerganovllm : add llm_build_context (#3881)
2023-11-01 bandoticommon : allow caller to handle help/argument exception...
2023-11-01 staviqlog : make generating separate log files optional ...
2023-11-01 l3utterflysampling : null grammar field after reset (#3885)
2023-11-01 Georgi Gerganovggml : fix UNUSED macro (#3762)
2023-11-01 Andrew Godfreyfinetune : add -ngl parameter (#3762)
2023-11-01 Georgi Gerganovscripts : add server-llm.sh (#3868)
2023-11-01 Adrian Heskethserver : re-enable completion and embedded at the same...
2023-11-01 Georgi Gerganovllama : refactor graph build code (#3837)
2023-10-31 kalomazesamplers : Min-P sampler implementation [alternative...
2023-10-31 Tungsten842flake.nix: fix for rocm 5.7 (#3853)
2023-10-30 Georgi Gerganovggml : move FP16 <-> FP32 code to ggml-impl.h (#3861)
2023-10-29 KerfuffleExtend llama_kv_cache_seq_rm to allow matching any...
2023-10-29 cebtenzzremake : remove unnecessary dependency on build-info...
2023-10-29 Georgi Gerganovllama : fix kv shift bug (#3835)
2023-10-29 Georgi Gerganovggml : quantization refactoring (#3833)
2023-10-28 Erik Scholzflake : update flake.lock for newer transformers versio...
2023-10-28 Aarni Koskelametal : try cwd for ggml-metal.metal if bundle lookup...
2023-10-28 Georgi Gerganovissues : change label from bug to bug-unconfirmed ...
2023-10-28 Georgi Gerganovconvert : ignore tokens if their IDs are within [0...
2023-10-28 Kerfufflellama : allow quantizing k-quants to fall back when...
2023-10-28 Georgi Gerganovllama : add option for greedy sampling with probs ...
2023-10-28 Henk Poleycommon : print that one line of the syntax help *also...
2023-10-28 Georgi Gerganovstarcoder : add GPU offloading (#3827)
2023-10-27 Kerfufflespeculative : ensure draft and target model vocab match...
2023-10-27 cebtenzzrellama : correctly report GGUFv3 format (#3818)
2023-10-27 Thibault Terrassonsimple : fix batch handling (#3803)
2023-10-27 Georgi Gerganovcuda : improve text-generation and batched decoding...
2023-10-26 Georgi Gerganovserver : do not release slot on image input (#3798)
2023-10-25 Georgi Gerganovbatched-bench : print params at start
2023-10-25 Georgi Gerganovlog : disable pid in log filenames
2023-10-24 cebtenzzreserver : add parameter -tb N, --threads-batch N (#3584...
2023-10-24 Georgi Gerganovserver : do not block system prompt update (#3767)
2023-10-24 Georgi Gerganovsync : ggml (conv ops + cuda MSVC fixes) (#3765)
2023-10-24 John Smithcmake : add missed dependencies (#3763)
2023-10-24 Georgi Gerganovcuda : add batched cuBLAS GEMM for faster attention...
2023-10-24 GalunidAdd more tokenizer tests (#3742)
2023-10-24 Georgi Gerganovmetal : handle ggml_scale for n%4 != 0 (close #3754)
2023-10-23 Georgi GerganovRevert "make : add optional CUDA_NATIVE_ARCH (#2482)"
2023-10-23 M. Yusuf Sarıgözissues : separate bug and enhancement template + no...
2023-10-23 GalunidUpdate special token handling in conversion scripts...
2023-10-23 Marcus Dunnllama : remove token functions with `context` args...
2023-10-23 GalunidFix baichuan convert script not detecing model (#3739)
2023-10-22 Alexmake : add optional CUDA_NATIVE_ARCH (#2482)
2023-10-22 Georgi Gerganovserver : parallel decoding and multimodal (#3677)
2023-10-22 goerchAdd test for MPT tokenization (#3728)
2023-10-22 Ian Scrivenerreadme : remove unsupported node.js library (#3703)
2023-10-22 Kerfufflellama : validate special token ids are in range when...
2023-10-22 vvhg1main : escape prompt for cfg_negative_prompt and consec...
2023-10-22 Georgi Gerganovbatched : add len CLI argument
2023-10-20 shibe2CLBlast: Add outer loops over src0 for broadcasting...
2023-10-20 Georgi Gerganovsampling : refactor init to use llama_sampling_params...
2023-10-20 Qin Yue Chengguf : support big endian platform (#3552)
2023-10-20 Georgi Gerganovserver : fix uninitialized sampling context (close...
2023-10-20 Herman Semenovggml : fix rope + llama minor optimizations (#3560)
2023-10-20 cebtenzzreconvert : restore compat with old Falcon models (#3680)
2023-10-19 M. Yusuf Sarıgözmultimodal : add BakLLaVA conversion support (#3682)
2023-10-19 M. Yusuf Sarıgözllava : avoid segfault in case of non-existent mmproj...
2023-10-18 Georgi Gerganovreadme : update hot topics
2023-10-18 Georgi Gerganovspeculative : bug fixes
2023-10-18 Georgi Gerganovspeculative : add tree-based sampling example (#3624)
2023-10-18 Jhen-Jie Hongmetal : implement q5_0 and q5_1 kernels (#3648)
2023-10-18 shibe2opencl : fix element-wise multiplication (#3656)
2023-10-17 slarenfix embeddings when using CUDA (#3657)
2023-10-17 Georgi Gerganovllama : avoid fprintf in favor of LLAMA_LOG (#3538)
2023-10-17 BarfingLemursreadme : update hot-topics & models, detail windows...
next