ggml-org/llama.cpp

Pull Requests Commits

kv-cache : avoid modifying recurrent cells when setting inputs

compilade committed 203 days ago

31296394

docs : Note about necessity of having libcurl installed for standard build. (#13945)

jpodivin committed 203 days ago

Verified b3a89c3d

server: allow unclosed thinking tags (#13931)

ochafik committed 203 days ago

Verified e15898d1

llama : deprecate explicit kv_self defrag/update calls (#13921)

ggerganov committed 203 days ago

Verified 803f8baf

llama : use n_swa + n_ubatch cells for SWA cache (#13833)

ggerganov committed 203 days ago

Verified 3600cc28

webui : Replace alert and confirm with custom modals. (#13711)

igardev committed 203 days ago

Verified c7e0a205

llama : auto-batch preparation (#13845)

ggerganov committed 203 days ago

Verified 3f55f781

mtmd : drop `_shared` from `libmtmd` name, merge helpers into libmtmd (⚠️ breaking change) (#13917)

ngxson committed 203 days ago

Verified 51fa76f1

kv-cache : refactor + add llama_memory_state_i (#13746)

ggerganov committed 203 days ago

Verified 12d0188c

CUDA: add a prop in ggml_cuda_device_infor for distinguish iGPU or dGPU in cuda (#13856) (#13895)

Yangxiaoz committed 203 days ago

Verified eb394993

CUDA: fix typo in FlashAttention code (#13926)

JohannesGaessler committed 204 days ago

Verified e562eece

sched : avoid changing cur_copy when a graph is already allocated (#13922)

slaren committed 204 days ago

Verified b47ab7b8

parallel : increase the variability of the prompt lengths (#13927)

ggerganov committed 204 days ago

Verified dd665cc9

cuda : prevent using split buffers with 3d/4d matrices (#13919)

slaren committed 204 days ago

Verified df0c0c7d

SYCL: Add mrope kernel (#13755)

qnixsynapse committed 204 days ago

Verified b49a8ff9

sync : vendor (#13901)

ggerganov committed 204 days ago

Verified 53f92507

convert : fix rwkv bos/eos token (#13844)

CISC committed 204 days ago

Verified db38704f

convert : allow partial update to the chkhsh pre-tokenizer list (#13847)

ngxson committed 204 days ago

Verified 07e4351c

llama : add support for DistilBert (#13907)

huydt84 committed 204 days ago

Verified 291f2b69

llama : use llm_build_granite for minicpm (#13911)

zkh2016 committed 204 days ago

Verified 2c90da4c

cmake: Guard GGML_CPU_ALL_VARIANTS by architecture (#13890)

ckastner committed 205 days ago

Verified ec9e0301

llama : add support for jina-reranker-v2 (#13900)

CISC committed 205 days ago

Verified e83ba3e4

gguf-py : add support for sub_type (in arrays) in GGUFWriter add_key_value method (#13561)

CISC committed 205 days ago

Verified 2b131621

arm64: optimize q4_k_q8_k kernel with i8mm (#13886)

cyb70289 committed 205 days ago

Verified 54a2c7a8

cmake: Factor out CPU architecture detection (#13883)

ckastner committed 205 days ago

Verified 21fcc21a

ggml: aarch64: Implement SVE F32 kernels for Mamba Sequential Scan Algorithm (#13882)

vineelabhinav committed 205 days ago

Verified dd8ba934

tests : remove json.hpp from a test (#13880)

ggerganov committed 205 days ago

Verified 66c92061

convert : workaround for AutoConfig dummy labels (#13881)

CISC committed 205 days ago

Verified 5ca82fc1

llama : add RobertaForSequenceClassification reranker support (#13875)

CISC committed 205 days ago

Verified 6385b843

ggml: aarch64: Implement SVE F32 kernels for vector functions (#13843)

vineelabhinav committed 205 days ago

Verified 1b8fb815

Newer Older