ggml-org/llama.cpp

Pull Requests Commits

graph : constant topology for tokens/embeddings inputs

ggerganov committed 11 days ago

Verified bf3f12df

graph : reduce topology branching

ggerganov committed 11 days ago

Verified 4ed59dc2

rpc : use unordered_map::reserve and emplace (#18513)

struct committed 11 days ago

Verified c6f0e832

cuda : fix copy of large tensors (ggml_nbytes <= INT_MAX assertion) (#18433)

Meet91721 committed 11 days ago

Verified e86f3c22

model : remove modern-bert iswa template (#18529)

CISC committed 11 days ago

Verified 169ee68f

model: support youtu-vl model (#18479)

f291400 committed 11 days ago

Verified ced765be

Add conversion support for IQuestCoderForCausalLM (#18524)

pwilkin committed 11 days ago

Verified 3ccccc83

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

o7si committed 11 days ago

Verified d0a6a314

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

o7si committed 11 days ago

Verified 2b2afade

model: add Solar Open model (#18511)

HelloKS committed 11 days ago

Verified f4f50192

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

Anri-Lombard committed 12 days ago

Verified d5574c91

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

am17an committed 12 days ago

Verified 26831bde

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

jeffbolznv committed 12 days ago

Verified be47fb92

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

triplenom committed 12 days ago

Verified 9e10bd2e

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

Anri-Lombard committed 12 days ago

Verified 4cd162a1

ggerganov committed 12 days ago

13814eb3

ggml : bump version to 0.9.5 (ggml/1410)

ggerganov committed 12 days ago

54f67b9b

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

Anri-Lombard committed 13 days ago

Verified 33ded988

convert : lint fix (#18507)

CISC committed 13 days ago

Verified 0db81098

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

Henry147147 committed 13 days ago

Verified 9b8329de

metal : add count_equal op (#18314)

gatbontonpc committed 13 days ago

Verified 9a6369bb

CUDA: fix KQ max calculation (#18487)

JohannesGaessler committed 13 days ago

Verified ecc343de

metal : remove BF16 x F16 kernels (#18456)

ggerganov committed 13 days ago

Verified 01ade96e

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

am17an committed 13 days ago

Verified 7bcaf815

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

rrsathe committed 13 days ago

Verified c8a37980

docker : add CUDA 13.1 image build (#18441)

CISC committed 13 days ago

Verified 4849661d

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

louwers committed 13 days ago

Verified 6e0c8cbc

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

aldehir committed 13 days ago

Verified 0f89d2ec

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

danbev committed 13 days ago

Verified ac1d0eb7

lora: count lora nodes in graph_max_nodes (#18469)

ngxson committed 14 days ago

Verified cd78e57c

Older