ngxson/llama.cpp

Pull Requests Commits

Merge branch 'master' into xsn/server_more_tests

ngxson committed 1 year ago

9864e0d2

sycl : Reroute permuted mul_mats through oneMKL (#10408)

Alberto Cabrera Pérez committed 1 year ago

Verified 266b8519

CANN: RoPE operator optimization (#10563)

noemotiovon committed 1 year ago

Verified 938f6087

vulkan: get the first command buffer submitted sooner (#10499)

jeffbolznv committed 1 year ago

Verified f095a649

llava: return false instead of exit (#10546)

tinglou committed 1 year ago

Verified 678d7994

ggml : remove redundant copyright notice + update authors

ggerganov committed 1 year ago

Verified dc223440

llama : add missing model types

ggerganov committed 1 year ago

Verified 4c0a95b1

server : (tests) don't use thread for capturing stdout/stderr, bump openai client library (#10568)

ngxson committed 1 year ago

Verified 6c595676

common: fix warning message when no GPU found (#10564)

JohannesGaessler committed 1 year ago

Verified 89071931

add invalid cases

ngxson committed 1 year ago

879c5ebd

docs: fix outdated usage of llama-simple (#10565)

rand-fly committed 1 year ago

Verified 7281cf13

ci : fix tag name in cuda and hip releases (#10566)

slaren committed 1 year ago

Verified e90688ed

add test speculative

ngxson committed 1 year ago

8aaf69a3

server : add split model test

ngxson committed 1 year ago

ac404be2

ggml : fix row condition for i8mm kernels (#10561)

ggerganov committed 1 year ago

Verified 76b27d29

cmake : fix ARM feature detection (#10543)

ggerganov committed 1 year ago

Verified eea986f2

ggml-cpu: support IQ4_NL_4_4 by runtime repack (#10541)

FanShupei committed 1 year ago

Verified c202cef1

kompute : improve backend to pass test_backend_ops (#10542)

slp committed 1 year ago

Verified 2025fa67

CANN: Update cann.md to display correctly in CLion (#10538)

HRXWEB committed 1 year ago

Verified c6bc7395

CANN: Fix SOC_TYPE compile bug (#10519)

leo-pony committed 1 year ago

Verified 605fa66c

CANN: ROPE operator optimization (#10540)

noemotiovon committed 1 year ago

Verified b7420131

common : fix duplicated file name with hf_repo and hf_file (#10550)

ngxson committed 1 year ago

Verified 9f912511

Add some minimal optimizations for CDNA (#10498)

IMbackK committed 1 year ago

Verified 3ad5451f

ci : faster CUDA toolkit installation method and use ccache (#10537)

slaren committed 1 year ago

Verified 46c69e0e

metal : fix group_norm support condition (#0)

ggerganov committed 1 year ago

Verified 9e2301f4

ggerganov committed 1 year ago

Verified fee824a1

Do not include arm_neon.h when compiling CUDA code (ggml/1028)

frankier committed 1 year ago

Verified 9150f8fe

vulkan: define all quant data structures in types.comp (#10440)

jeffbolznv committed 1 year ago

Verified c31ed2ab

vulkan: Handle GPUs with less shared memory (#10468)

jeffbolznv committed 1 year ago

Verified 5b3466be

vulkan: further optimize q5_k mul_mat_vec (#10479)

jeffbolznv committed 1 year ago

Verified 249a7902

Older