ngxson/llama.cpp

Pull Requests Commits

ngxson committed 1 year ago

e9d251ef

ngxson committed 1 year ago

0e728796

ngxson committed 1 year ago

f24460c4

ngxson committed 1 year ago

e8a4c814

launch_slot_with_task

ngxson committed 1 year ago

c240bd70

update completion.js

ngxson committed 1 year ago

02e49353

correct coding style

ngxson committed 1 year ago

4840c4e6

ngxson committed 1 year ago

0f641ebd

revert limit max n_predict

ngxson committed 1 year ago

400d4e63

fix api key test case

ngxson committed 1 year ago

c15f5a6e

Merge branch 'master' into xsn/better_error

ngxson committed 1 year ago

d2e1411c

server: do not crash on grammar error

ngxson committed 1 year ago

de81c22a

server: format error to json

ngxson committed 1 year ago

ba9c3e31

ggml : remove old quantization functions (#5942)

ggerganov committed 1 year ago

Verified 5b097973

server : clarify some items in the readme (#5957)

ggerganov committed 1 year ago

Verified 97c09585

server : normalize embeddings (#5956)

redlion0929 committed 1 year ago

Verified fb215c38

tests : gitignore ggml-common.h

ggerganov committed 1 year ago

Verified 2c4f566c

server : fix passing prompt as tokens (#5955)

z80maniac committed 1 year ago

Verified 0db32bea

ggml : add ggml-common.h to deduplicate shared code (#5940)

ggerganov committed 1 year ago

Verified 8a3012a4

server : simplify logic for empty prompts (#5953)

ggerganov committed 1 year ago

Verified 9674aaf3

Server: reorganize some http logic (#5939)

ngxson committed 1 year ago

Verified 950ba1ab

server : add SSL support (#5926)

gabe-l-hart committed 1 year ago

Verified e1fa9569

server: tests: add truncated prompt tests, better kv cache size (#5933)

phymbert committed 1 year ago

Verified fd72d2d2

llama : support Mamba Selective State Space Models (#5328)

compilade committed 1 year ago

Verified c2101a2e

llama : fix quantization of shared token_embd (#5944)

compilade committed 1 year ago

Verified 515f7d0d

server: metrics: add llamacpp:prompt_seconds_total and llamacpp:tokens_predicted_seconds_total, reset bucket only on /metrics. Fix values cast to int. Add Process-Start-Time-Unix header. (#5937)

phymbert committed 1 year ago

Verified 76e86882

llama : assume tied weights if lm_head/output weights is missing (#5824)

dmahurin committed 1 year ago

Verified e457fb35

server : fix EOS token detection with disabled cache (#5938)

ggerganov committed 1 year ago

Verified af37fd8b

log : fix MSVC compile errors (#5643)

uextm committed 1 year ago

Verified 581ed5c4

llama-bench : add embeddings option (#5924)

ggerganov committed 1 year ago

Verified 6cdabe65

Older