ngxson/llama.cpp

Pull Requests Commits

normalize newlines

CISC committed 71 days ago

Verified ca8d4ca5

check if empty first

CISC committed 71 days ago

Verified 8a88770f

hopefully fix windows

CISC committed 71 days ago

Verified c9a94e77

loosen ensure_vals max count condition, add not impl for map(int)

ngxson committed 71 days ago

e07af2bd

fix array.map()

ngxson committed 71 days ago

25dac2e0

add fuzz func_args, refactor to be safer

ngxson committed 71 days ago

350d87da

CISC committed 71 days ago

Verified 8e1e6ae7

CISC committed 71 days ago

Verified acaf0170

CISC committed 71 days ago

Verified 70d9d9c5

ngxson committed 71 days ago

10a987ae

Merge branch 'master' into xsn/jinja_vm

ngxson committed 71 days ago

78a0112d

fix chat_template_kwargs

ngxson committed 71 days ago

a959ff8b

ngxson committed 71 days ago

abcd7766

ngxson committed 71 days ago

f475f5bc

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

ggerganov committed 71 days ago

Verified 39173bca

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

JohannesGaessler committed 71 days ago

Verified 5c662d21

jinja : static analysis + header cleanup + minor fixes

ggerganov committed 71 days ago

Verified 50568649

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

shalinib-ibm committed 71 days ago

Verified 8cc0ba95

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

ngxson committed 71 days ago

Verified a7e6ddb8

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

CISC committed 71 days ago

Verified 2a131801

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

angt committed 71 days ago

Verified ec997b4f

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

max-krasnyansky committed 72 days ago

Verified cff777f2

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

ORippler committed 72 days ago

Verified 36f01324

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

pwilkin committed 72 days ago

Verified d98b5481

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

nuxlear committed 72 days ago

Verified 8fb71755

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

angt committed 72 days ago

Verified 516a4ca9

vulkan: Check maxStorageBufferRange in supports_op (#18709)

jeffbolznv committed 72 days ago

Verified 3e4bb296

llama-model: fix unfortunate typo (#18832)

am17an committed 72 days ago

Verified 47f96124

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

danbev committed 72 days ago

Verified 01cbdfd7

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

0cc4m committed 72 days ago

Verified 635ef78e

Older