ggml-org/llama.cpp

Pull Requests Commits

disable on windows

CISC committed 1 year ago

Verified d3a2eb59

revert build changes

CISC committed 1 year ago

Verified 7210ebe2

add arch to matrix

CISC committed 1 year ago

Verified 05f94a0e

download in batches

CISC committed 1 year ago

Verified f9a27178

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

CISC committed 1 year ago

Verified de8ec134

copy curl dll for tests

CISC committed 1 year ago

Verified 8e1125a8

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

jpodivin committed 1 year ago

Verified b3a89c3d

server: allow unclosed thinking tags (#13931)

ochafik committed 1 year ago

Verified e15898d1

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

ggerganov committed 1 year ago

Verified 803f8baf

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

ggerganov committed 1 year ago

Verified 3600cc28

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

igardev committed 1 year ago

Verified c7e0a205

llama : auto-batch preparation (#13845)

ggerganov committed 1 year ago

Verified 3f55f781

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

ngxson committed 1 year ago

Verified 51fa76f1

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

ggerganov committed 1 year ago

Verified 12d0188c

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

Yangxiaoz committed 1 year ago

Verified eb394993

CUDA: fix typo in FlashAttention code (#13926)

JohannesGaessler committed 1 year ago

Verified e562eece

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

slaren committed 1 year ago

Verified b47ab7b8

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

ggerganov committed 1 year ago

Verified dd665cc9

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

slaren committed 1 year ago

Verified df0c0c7d

SYCL: Add mrope kernel (#13755)

qnixsynapse committed 1 year ago

Verified b49a8ff9

sync : vendor (#13901)

ggerganov committed 1 year ago

Verified 53f92507

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

CISC committed 1 year ago

Verified db38704f

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

ngxson committed 1 year ago

Verified 07e4351c

llama : add support for DistilBert (#13907)

huydt84 committed 1 year ago

Verified 291f2b69

windows builds adds build type to runtime output

CISC committed 1 year ago

Verified 4b4843ad

llama : use llm_build_granite for minicpm (#13911)

zkh2016 committed 1 year ago

Verified 2c90da4c

cmake: Guard GGML_CPU_ALL_VARIANTS by architecture (#13890)

ckastner committed 1 year ago

Verified ec9e0301

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

CISC committed 1 year ago

Verified e83ba3e4

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

CISC committed 1 year ago

Verified 2b131621

arm64: optimize q4_k_q8_k kernel with i8mm (#13886)

cyb70289 committed 1 year ago

Verified 54a2c7a8

Older