ngxson/llama.cpp

Pull Requests Commits

server: improve slots scheduling for n_cmpl (#18789)

ngxson committed 6 days ago

Verified a04c2b06

context : reserve new scheduler when graph topology changes (#18547)

ggerganov committed 6 days ago

Verified 39173bca

CUDA: fix allignment on register spill for FA (#18815)

JohannesGaessler committed 6 days ago

Verified 5c662d21

ggml-cpu: optimize ggml_vec_dot_bf16 for Power9 (#18837)

shalinib-ibm committed 6 days ago

Verified 8cc0ba95

lora: make sure model keep track of associated adapters (#18490)

ngxson committed 6 days ago

Verified a7e6ddb8

model-loader : support bool array sliding window pattern (#18850)

CISC committed 6 days ago

Verified 2a131801

tests : download models only when running ctest (#18843)

angt committed 6 days ago

Verified ec997b4f

hexagon: support for OP_CPY, host buffers now optional, hvx-utils refactoring and optimizations (#18822)

max-krasnyansky committed 6 days ago

Verified cff777f2

CUDA: Factor out and re-use `block_reduce` function (#18785)

ORippler committed 6 days ago

Verified 36f01324

Restore clip's cb() to its rightful glory - extract common debugging elements in llama (#17914)

pwilkin committed 6 days ago

Verified d98b5481

model : clean up and fix EXAONE-MoE configuration (#18840)

nuxlear committed 6 days ago

Verified 8fb71755

refactor : remove libcurl, use OpenSSL when available (#18828)

angt committed 6 days ago

Verified 516a4ca9

vulkan: Check maxStorageBufferRange in supports_op (#18709)

jeffbolznv committed 7 days ago

Verified 3e4bb296

llama-model: fix unfortunate typo (#18832)

am17an committed 7 days ago

Verified 47f96124

CUDA : fix typo in clang pragma comment [no ci] (#18830)

danbev committed 7 days ago

Verified 01cbdfd7

vulkan: work around Intel fp16 bug in mmq (#18814)

0cc4m committed 7 days ago

Verified 635ef78e

ggml-metal: do not copy headers for embedded, use current binary dir for embedded (#18705)

DaAwesomeP committed 7 days ago

Verified 7d587e55

mmap: add Haiku support by skipping RLIMIT_MEMLOCK check (#18819)

danielbenjaminsson committed 7 days ago

Verified d34aa071

ci, tests : use cmake to download models and remove libcurl dependency (#18791)

angt committed 7 days ago

Verified f709c7a3

llama : print_info alignment fix (#18708)

ddh0 committed 7 days ago

Verified 6e36299b

model : add EXAONE MoE (#18543)

nuxlear committed 7 days ago

Verified 60591f01

vocab : fix attribute overrides for harmony (#18806)

ggerganov committed 8 days ago

Verified e4832e3a

llama-mmap: fix direct-io loading fallback EOF exception (#18801)

0cc4m committed 8 days ago

Verified 960e5e3b

model-conversion : remove -c 0 from model card template [no ci] (#18807)

danbev committed 8 days ago

Verified 20ca2e12

HIP: add fattn-mma-f16 for RDNA4 (#18481)

zhang-hui-yulo committed 8 days ago

Verified ea4a321f

doc: ban AI-generated PR descriptions [no ci] (#18765)

JohannesGaessler committed 8 days ago

Verified c1e79e61

mtmd: fix use_non_causal being reported incorrectly (#18793)

ngxson committed 8 days ago

Verified e047f9ee

CUDA : fix unused argument when USE_CUDA_GRAPH=OFF (#18800)

ggerganov committed 8 days ago

Verified 0a57271a

graph : clean up t5 input builders (#18795)

gabe-l-hart committed 8 days ago

Verified 076b0faf

llama-bench: add direct_io parameter (#18778)

0cc4m committed 8 days ago

Verified db79dc06

Newer Older