ggerganov/llama.cpp

Pull Requests Commits

model : add support for JinaBertModel with non-gated ffn (#18475)

o7si committed 16 days ago

Verified d0a6a314

convert : fix encoding of WPM vocab for BERT models (#18500)

o7si committed 16 days ago

Verified 2b2afade

model: add Solar Open model (#18511)

HelloKS committed 16 days ago

Verified f4f50192

webui: fix code copy stripping XML/HTML tags (#18518)

Anri-Lombard committed 16 days ago

Verified d5574c91

ggml-cuda: remove unneccesary prints on ggml_cuda_init (#18502)

am17an committed 16 days ago

Verified 26831bde

vulkan: extend topk_moe to handle sigmoid w/exp_probs_b for nemotron (#18295)

jeffbolznv committed 17 days ago

Verified be47fb92

llama: handle short reads in direct I/O path (#18504)

triplenom committed 17 days ago

Verified 9e10bd2e

chat: make tool description and parameters optional per OpenAI spec (#18478)

Anri-Lombard committed 17 days ago

Verified 4cd162a1

ggerganov committed 17 days ago

13814eb3

ggml : bump version to 0.9.5 (ggml/1410)

ggerganov committed 17 days ago

54f67b9b

quantize: prevent input/output file collision (#18451)

Anri-Lombard committed 17 days ago

Verified 33ded988

convert : lint fix (#18507)

CISC committed 17 days ago

Verified 0db81098

mtmd : Adding support for Nvidia Music Flamingo Model (#18470)

Henry147147 committed 17 days ago

Verified 9b8329de

metal : add count_equal op (#18314)

gatbontonpc committed 18 days ago

Verified 9a6369bb

CUDA: fix KQ max calculation (#18487)

JohannesGaessler committed 18 days ago

Verified ecc343de

metal : remove BF16 x F16 kernels (#18456)

ggerganov committed 18 days ago

Verified 01ade96e

sycl: add newline at the end of CMakeLists.txt (#18503)

am17an committed 18 days ago

Verified 7bcaf815

Work around broken IntelSYCLConfig.cmake in Intel oneAPI 2025.x (#18345)

rrsathe committed 18 days ago

Verified c8a37980

docker : add CUDA 13.1 image build (#18441)

CISC committed 18 days ago

Verified 4849661d

docs : document that JSON Schema is not available to model when using response_format (#18492)

louwers committed 18 days ago

Verified 6e0c8cbc

common : default content to an empty string (#18485)

aldehir committed 18 days ago

Verified 0f89d2ec

llama : fix typo in comment in llama-kv-cache.h [no ci] (#18489)

danbev committed 18 days ago

Verified ac1d0eb7

lora: count lora nodes in graph_max_nodes (#18469)

ngxson committed 18 days ago

Verified cd78e57c

sampling: reuse token data buffer in llama_sampler_sample (#18365)

JayZenith committed 18 days ago

Verified c32fa21d

server: fix files built redundantly (#18474)

jeffbolznv committed 18 days ago

Verified f14f4e42

kleidiai: add and integrate SVE 256-bit vector-length kernel (#18458)

chaxu01 committed 18 days ago

Verified 2d6c00a9

CUDA: add log line when mxfp4 acceleration is used (#18483)

am17an committed 18 days ago

Verified d77d7c5c

model-conversion : use CONVERTED_MODEL for compare-embeddings (#18461)

danbev committed 18 days ago

Verified a864fb1c

webui: fix prompt progress ETA calculation (#18468)

ngxson committed 19 days ago

Verified 51a48720

Webui/prompt processing progress (#18300)

ServeurpersoCom committed 19 days ago

Verified c9a3b40d

Newer Older