ochafik/llama.cpp

Pull Requests Commits

Update common/arg.cpp

ochafik committed 1 year ago

Verified 153a0b87

rm OFFLINE from server tests (LLAMA_OFFLINE=1 works)

ochafik committed 1 year ago

2eb3c699

ochafik committed 1 year ago

f8074b9d

refactor as suggested

ochafik committed 1 year ago

0818ab21

Merge remote-tracking branch 'origin/master' into offline

ochafik committed 1 year ago

7f7d8597

scripts : add option to compare commits in Debug (#13806)

ggerganov committed 1 year ago

Verified a26c4cc1

cuda : avoid cuGetErrorString (#13791)

ggerganov committed 1 year ago

Verified 4265a87b

ochafik committed 1 year ago

31ea1b86

SYCL: Add non contiguous support in RMS_NORM and NORM kernels (#13611)

qnixsynapse committed 1 year ago

Verified 6f180b91

ochafik committed 1 year ago

f7439cbf

fix signatures of ifdef'd function fallbacks

ochafik committed 1 year ago

bb213c6a

Merge remote-tracking branch 'origin/master' into offline

ochafik committed 1 year ago

40712c5a

server: fix streaming crashes (#13786)

ochafik committed 1 year ago

Verified 03f582ae

examples/training: Fix file name in README (#13803)

standby24x7 committed 1 year ago

Verified 88c125f2

`server`: fix format of streamed tool call deltas (diff name, fix id location) (#13800)

ochafik committed 1 year ago

Verified d74e94c1

server: fix regression on streamed non-chat completion w/ stops (#13785)

ochafik committed 1 year ago

Verified f13847cf

use LLAMA_OFFLINE env for --offline

ochafik committed 1 year ago

273a7285

examples : allow extracting embeddings from decoder contexts (#13797)

ggerganov committed 1 year ago

Verified 79c137f7

llama : clarify deprecation message (#13794)

ggerganov committed 1 year ago

Verified 22229314

sycl: Add more debug prints (#13640)

Rbiessy committed 1 year ago

Verified 9012eb9b

vulkan: mark IM2COL as supporting non-contig (#13783)

jeffbolznv committed 1 year ago

Verified fef693dc

CANN: Add the basic supports of Flash Attention kernel (#13627)

shibizhao committed 1 year ago

Verified 2d38b6e4

server: --offline mode

ochafik committed 1 year ago

21b59887

`server`: add `--reasoning-budget 0` to disable thinking (incl. qwen3 w/ enable_thinking:false) (#13771)

ochafik committed 1 year ago

Verified e121edc4

webui : bump max upload file size to 500MB (#13779)

ngxson committed 1 year ago

Verified 2f099b51

tests : improve UGM tokenizer test coverage (#13773)

CISC committed 1 year ago

Verified aa50ba46

kv-cache : rework kv_cell (#13706)

ggerganov committed 1 year ago

Verified de2ef53a

rpc : Fix build on OpenBSD (#13541)

percypiper committed 1 year ago

Verified c508256d

mtmd : add support for Qwen2-Audio and SeaLLM-Audio (#13760)

ngxson committed 1 year ago

Verified 40aaa8a4

docs : add Moondream2 pre-quantized link (#13745)

ddpasa committed 1 year ago

Verified a08c1d28

Older