ggerganov/llama.cpp

Pull Requests Commits

tts : fix n_ubatch + make WavTokenizer cache-less

ggerganov committed 201 days ago

Verified 108d484a

mtmd : add ultravox audio input (#13623)

ngxson committed 201 days ago

Verified 797990c4

common: Include torch package for s390x (#13699)

taronaeo committed 201 days ago

Verified ab863357

server : pad small embedding batches (#13692)

ggerganov committed 201 days ago

Verified cc74d5be

gguf-py : correct charsmap parameter typing (#13701)

CISC committed 201 days ago

Verified 5be24af7

sycl : Remove waits from function calls (#13702)

s-Nick committed 201 days ago

Verified d394a9ae

SYCL: Avoid using with SYCL-Graph for unsupported nodes (#13587)

Ewan Crawford committed 201 days ago

Verified 6b56a646

opencl: Add support for multiple devices (#12622)

Henry Linjamäki committed 202 days ago

Verified a4e8912d

opencl: fix couple crashes (#12795)

Henry Linjamäki committed 202 days ago

Verified edbf42ed

releases : build CPU backend separately (windows) (#13642)

slaren committed 202 days ago

Verified d643bb2c

hparams : support models for which all layers use SWA (#13682)

ggerganov committed 202 days ago

Verified 8e186ef0

server : improve error reporting (#13680)

ggerganov committed 202 days ago

Verified 5fbfe384

convert : add qwen2vl support for unsloth merges (#13686)

antichristHater committed 202 days ago

Verified c76532e7

examples : switch retrieval to llama_encode (#13685)

CISC committed 202 days ago

Verified 2aa777d8

gguf-py : display the invalid gguf type (#13687)

emmanuel-ferdman committed 202 days ago

Verified eb0f5c28

ggml : add ggml_gelu_erf() (#13667)

ngxson committed 202 days ago

Verified cf4cb59e

server : Add the endpoints /api/tags and /api/chat (#13659)

R-Dson committed 202 days ago

Verified 0d5c7421

server : fix first message identification (#13634)

doringeman committed 202 days ago

Verified 42158ae2

kv-cache : simplify the interface (#13660)

ggerganov committed 202 days ago

Verified 797f2ac0

model : disable SWA for Phi models (#13676)

ggerganov committed 202 days ago

Verified b44890df

musa: Upgrade MUSA SDK version to rc4.0.1 and use mudnn::Unary::IDENTITY op to accelerate D2D memory copy (#13647)

yeahdongcn committed 202 days ago

Verified 33983057

vulkan: fix warnings (#13626)

netrunnereve committed 203 days ago

Verified fb1cab20

mtmd-helper : bug fix to token batching in mtmd (#13650)

l3utterfly committed 203 days ago

Verified b7a17463

model : fix llama4 graph (#13663)

ggerganov committed 203 days ago

Verified be023969

llama : remove llama_kv_cache_view API + remove deprecated (#13653)

ggerganov committed 203 days ago

Verified a4090d11

CUDA: skip fully masked-out KV in FA vec kernel (#13584)

JohannesGaessler committed 203 days ago

Verified b69f1647

tests : avoid github urls due to throttling (#13654)

CISC committed 203 days ago

Verified 759e37b0

sycl: disable reorder for sycl mulmat (#13536)

sgeor255 committed 203 days ago

Verified 4245e622

Set GLM4 blk.*.attn_output.weight, kqv_out-* matmul to GGML_PREC_F32 to fix infinity values in output (#13639)

0cc4m committed 203 days ago

Verified c9c64dee

metal : fix typo in FA kernel comments (#13651)

ggerganov committed 203 days ago

Verified c00a2634

Older