ggerganov/llama.cpp

Pull Requests Commits

server : fix LRU check (#14079)

ggerganov committed 90 days ago

Verified 87d34b38

sycl: Add reorder to Q6_K mmvq implementation (#13885)

s-Nick committed 90 days ago

Verified b460d16a

add geglu activation function (#14074)

huydt84 committed 90 days ago

Verified 91a8ee6a

CANN: Enable labeler for Ascend NPU (#13914)

shink committed 91 days ago

Verified 056eb745

cuda : fix buffer type check with integrated GPUs (#14069)

slaren committed 91 days ago

Verified 247e5c6e

ci: add LoongArch cross-compile build (#13944)

wojiushixiaobai committed 92 days ago

Verified 5787b5da

SYCL: Implement few same quantized type copy kernels (#13739)

qnixsynapse committed 92 days ago

Verified 228f34c9

llama : fix llama_model_chat_template with template name (LLM_KV with suffix) (#14050)

CISC committed 92 days ago

Verified 0974ad7a

llama : deprecate llama_kv_self_ API (#14030)

ggerganov committed 93 days ago

Verified 745aa531

context : fix SWA-related warning for multiple sequences (#14045)

ggerganov committed 93 days ago

Verified 487a5e04

llama : support multiple classifier outputs and labels (#13940)

CISC committed 93 days ago

Verified d17a809e

gguf-py : add add_classifier_output_labels method to writer (#14031)

CISC committed 94 days ago

Verified 1caae7fc

vulkan: Enable VK_KHR_cooperative_matrix extension for Intel Xe2 GPUs (#14001)

rillomas committed 94 days ago

Verified 669c13e0

ci: fix CUDA build failure on autodl cloud machines (#14005)

pockers21 committed 94 days ago

Verified 146b88e8

memory : migrate from llama_kv_cache to more generic llama_memory (#14006)

ggerganov committed 94 days ago

Verified 7f37b6cf

llama : allow using mmap without PrefetchVirtualMemory, apply GGML_WIN_VER to llama.cpp sources (#14013)

slaren committed 94 days ago

Verified 3a077146

readme : add badge (#13938)

Olexandr88 committed 94 days ago

Verified d01d112a

vocab : warn about missing mask token (#14022)

CISC committed 94 days ago

Verified 9f47fa57

context : fix pos_min initialization upon error decode (#14008)

ggerganov committed 94 days ago

Verified 9e31bec4

vulkan: automatically deduce size of push constants (#13936)

jeffbolznv committed 94 days ago

Verified 5a8ae305

ggml-vulkan: adds support for op CONV_TRANSPOSE_1D (#13813)

etasnadi committed 95 days ago

Verified 0d398442

kv-cache : refactor the update/defrag mechanism (#13988)

ggerganov committed 95 days ago

Verified 3e63a58e

ci : remove cuda 11.7 releases, switch runner to windows 2022 (#13997)

slaren committed 95 days ago

Verified 2589ad37

releases : use dl backend for linux release, remove arm64 linux release (#13996)

slaren committed 95 days ago

Verified 48254871

llama-graph : use ggml_repeat_4d (#13998)

ngxson committed 95 days ago

Verified 3ac67535

CUDA: fix FTZ in FA for Gemma 3 (#13991)

JohannesGaessler committed 95 days ago

Verified 0b4be4c4

kv-cache : fix unified::seq_rm to work with seq_id < 0 (#13985)

ggerganov committed 95 days ago

Verified e0e806f5

vulkan: fix warnings in perf logger querypool code (#13937)

jeffbolznv committed 96 days ago

Verified 7e00e60e

docs : add "Quick start" section for new users (#13862)

ngxson committed 96 days ago

Verified ea1431b0

opencl: add `backend_synchronize` (#13939)

lhez committed 97 days ago

Verified 71e74a3a

Newer Older