ggerganov/llama.cpp

Pull Requests Commits

disable on windows

CISC committed 96 days ago

Verified d3a2eb59

revert build changes

CISC committed 96 days ago

Verified 7210ebe2

add arch to matrix

CISC committed 96 days ago

Verified 05f94a0e

download in batches

CISC committed 96 days ago

Verified f9a27178

Merge branch 'master' into cisc/test-tokenizers-remote

CISC committed 96 days ago

Verified de8ec134

copy curl dll for tests

CISC committed 96 days ago

Verified 8e1125a8

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

jpodivin committed 96 days ago

Verified b3a89c3d

server: allow unclosed thinking tags (#13931)

ochafik committed 96 days ago

Verified e15898d1

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

ggerganov committed 96 days ago

Verified 803f8baf

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

ggerganov committed 96 days ago

Verified 3600cc28

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

igardev committed 96 days ago

Verified c7e0a205

llama : auto-batch preparation (#13845)

ggerganov committed 96 days ago

Verified 3f55f781

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

ngxson committed 96 days ago

Verified 51fa76f1

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

ggerganov committed 96 days ago

Verified 12d0188c

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

Yangxiaoz committed 96 days ago

Verified eb394993

CUDA: fix typo in FlashAttention code (#13926)

JohannesGaessler committed 97 days ago

Verified e562eece

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

slaren committed 97 days ago

Verified b47ab7b8

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

ggerganov committed 97 days ago

Verified dd665cc9

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

slaren committed 97 days ago

Verified df0c0c7d

SYCL: Add mrope kernel (#13755)

qnixsynapse committed 97 days ago

Verified b49a8ff9

sync : vendor (#13901)

ggerganov committed 97 days ago

Verified 53f92507

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

CISC committed 97 days ago

Verified db38704f

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

ngxson committed 97 days ago

Verified 07e4351c

llama : add support for DistilBert (#13907)

huydt84 committed 97 days ago

Verified 291f2b69

windows builds adds build type to runtime output

CISC committed 97 days ago

Verified 4b4843ad

llama : use llm_build_granite for minicpm (#13911)

zkh2016 committed 97 days ago

Verified 2c90da4c

cmake: Guard GGML_CPU_ALL_VARIANTS by architecture (#13890)

ckastner committed 98 days ago

Verified ec9e0301

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

CISC committed 98 days ago

Verified e83ba3e4

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

CISC committed 98 days ago

Verified 2b131621

arm64: optimize q4_k_q8_k kernel with i8mm (#13886)

cyb70289 committed 98 days ago

Verified 54a2c7a8

Older