ngxson/llama.cpp

Pull Requests Commits

add llm_graph_input_one

ngxson committed 248 days ago

9c98bab4

ngxson committed 248 days ago

08f7f2b0

llama : return mistral-v7-tekken as default template only (#14390)

CISC committed 248 days ago

Verified b2534622

metal : add special-case mat-vec mul for ne00 == 4 (#14385)

ggerganov committed 248 days ago

Verified e8215dbb

metal : batch rows copy in a single threadgroup (#14384)

ggerganov committed 248 days ago

Verified 5783ae43

docs: update s390x documentation + add faq (#14389)

taronaeo committed 248 days ago

Verified bf5bcd0b

musa: enable fp16 mma (all) and cublas on qy2 (#13842)

yeahdongcn committed 249 days ago

Verified 716301d1

ggml-cpu: enable IBM NNPA Vector Intrinsics (#14317)

taronaeo committed 249 days ago

Verified 60ef23d6

ggml : do not output unprintable characters on GGUF load failure (#14381)

CISC committed 249 days ago

Verified b193d530

sycl: GGML_SYCL_DISABLE_OPT on by default for all Intel Devices (#13973)

ShanoToni committed 249 days ago

Verified 2bf9d539

opencl: ref count `ggml_backend_opencl_context` and refactor profiling (#14254)

lhez committed 250 days ago

Verified 73e53dc8

batch : fix check for empty sequences in memory (#14364)

ggerganov committed 250 days ago

Verified 62af4642

cmake : use LLAMA_BUILD_NUMBER when defining LLAMA_INSTALL_VERSION (#14362)

mbaudier committed 250 days ago

Verified c148cf19

server : move no API key doc to /health (#14352)

pnb committed 250 days ago

Verified 1b809cee

main : honor --verbose-prompt on interactive prompts (#14350)

CISC committed 250 days ago

Verified abf24104

jinja : Add Mistral-Small-3.2-24B-Instruct-2506.jinja (#14349)

bartowski1182 committed 250 days ago

Verified 901e20bb

CUDA/HIP: optimize mmv paths taken for HIP devices (#14324)

IMbackK committed 251 days ago

Verified 0142961a

ci: add workflow for relocatable cmake package (#14346)

mtmcp committed 251 days ago

Verified ce82bd01

vulkan: update windows SDK in release.yml (#14344)

jeffbolznv committed 251 days ago

Verified bf2a99e3

llama : better rwkv chat template and add missing `inputs.use_jinja` setting (#14336)

MollySophia committed 251 days ago

Verified 72c6bc3f

CUDA: mul_mat_v support for batch sizes > 1 (#14262)

JohannesGaessler committed 251 days ago

Verified defe2158

kv-cells : fix tracking of seq_pos (#14339)

ggerganov committed 251 days ago

Verified 7b50d589

vulkan: update windows SDK in CI (#14334)

jeffbolznv committed 251 days ago

Verified 3a9457df

quantize : handle user-defined pruning of whole layers (blocks) (#13037)

EAddario committed 252 days ago

Verified fa4a9f2a

gguf-py : fix SpecialVocab parsing when post_processor is null (#14330)

CISC committed 252 days ago

Verified 238005c2

run : avoid double tokenization (#14327)

retr0reg committed 252 days ago

Verified 66aba7ac

examples : fix is_first logic for tokenization (#14329)

ggerganov committed 252 days ago

Verified f1f5e82d

HIP: enable vec fattn on RDNA4 (#14323)

IMbackK committed 252 days ago

Verified af3373f1

mtmd : fix Pixtral OOM with large images by capping image_size to 1024 (#14326)

yuiseki committed 252 days ago

Verified 5d5c066d

common : use std::string_view now that we target c++17 (#14319)

CISC committed 252 days ago

Verified 40bfa04c

Older