ngxson/llama.cpp

Pull Requests Commits

fix 3, 4, add 5, 6

ngxson committed 191 days ago

41806dc5

py quants: wip q4_k and q3_k quantize

ngxson committed 191 days ago

29ca8f25

metal : use less stack memory in FA kernel (#14088)

ggerganov committed 192 days ago

Verified 1f63e75f

kv-cache : fix shift and defrag logic (#14081)

ggerganov committed 192 days ago

Verified 40cbf571

llama : allow building all tests on windows when not using shared libs (#13980)

slaren committed 192 days ago

Verified 7f4fbe51

ggml-cpu : split arch-specific implementations (#13892)

xctan committed 192 days ago

Verified f470bc36

cuda : fix device sync on buffer clear (#14033)

slaren committed 192 days ago

Verified 8f47e25f

graph : fix geglu (#14077)

ggerganov committed 192 days ago

Verified 201b31dc

CANN: Simplify the environment variable setting(#13104)

bachelor-dou committed 192 days ago

Verified e21d2d4a

webui: fix sidebar being covered by main content (#14082)

yeahdongcn committed 192 days ago

Verified dc0623fd

server : fix LRU check (#14079)

ggerganov committed 192 days ago

Verified 87d34b38

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

s-Nick committed 192 days ago

Verified b460d16a

add geglu activation function (#14074)

huydt84 committed 192 days ago

Verified 91a8ee6a

CANN: Enable labeler for Ascend NPU (#13914)

shink committed 192 days ago

Verified 056eb745

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

slaren committed 193 days ago

Verified 247e5c6e

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

wojiushixiaobai committed 194 days ago

Verified 5787b5da

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

qnixsynapse committed 194 days ago

Verified 228f34c9

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

CISC committed 194 days ago

Verified 0974ad7a

llama : deprecate llama_kv_self_ API (#14030)

ggerganov committed 195 days ago

Verified 745aa531

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

ggerganov committed 195 days ago

Verified 487a5e04

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

CISC committed 195 days ago

Verified d17a809e

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

CISC committed 196 days ago

Verified 1caae7fc

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

rillomas committed 196 days ago

Verified 669c13e0

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

pockers21 committed 196 days ago

Verified 146b88e8

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

ggerganov committed 196 days ago

Verified 7f37b6cf

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

slaren committed 196 days ago

Verified 3a077146

readme : add badge (#13938)

Olexandr88 committed 196 days ago

Verified d01d112a

vocab : warn about missing mask token (#14022)

CISC committed 196 days ago

Verified 9f47fa57

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

ggerganov committed 196 days ago

Verified 9e31bec4

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

jeffbolznv committed 196 days ago

Verified 5a8ae305

Older