ggerganov/llama.cpp

Pull Requests Commits

metal : print more GPU info + disable mul_mm for MTLGPUFamiliy < Apple7

ggerganov committed 1 year ago

Verified 6b9554a7

ggerganov committed 1 year ago

Verified 545b0349

metal : indentations

ggerganov committed 1 year ago

Verified 8f6ad684

metal : rename kernels mul_mat_ to mul_mv_

ggerganov committed 1 year ago

Verified c6002248

metal : improve decoding speed for batches of 2-16

ggerganov committed 1 year ago

Verified 99ed03a2

quantize : fail fast on write errors (#3521)

cebtenzzre committed 1 year ago

Verified f1782c68

metal : support default.metallib load & reuse code for swift package (#3522)

jhen0409 committed 1 year ago

Verified c26765a0

llm : support Adept Persimmon 8B (#3410)

phillip-kravtsov committed 1 year ago

Verified 0e797c2f

Fix for #3454 (#3455)

goerch committed 1 year ago

Verified 3a716b4d

readme : update models, cuda + ppl instructions (#3510)

BarfingLemurs committed 1 year ago

Verified 1faaae8c

server : docs fix default values and add n_probs (#3506)

Mihaiii committed 1 year ago

Verified cb13d73a

kv cache slot search improvements (#3493)

KerfuffleV2 committed 1 year ago

Verified 9ca79d5c

prompts : fix editorconfig checks after #3416

ggerganov committed 1 year ago

Verified 0c731ca4

parallel : add option to load external prompt file (#3416)

pudepiedj committed 1 year ago

Verified a8777ad8

server : reuse llama_sample_token common util (#3494)

jhen0409 committed 1 year ago

Verified 97af49fa

llama : correct hparams comparison (#3446)

l3utterfly committed 1 year ago

Verified 16820a5a

ci : fix xcodebuild destinations (#3491)

jhen0409 committed 1 year ago

Verified 04b2f438

convert : update Falcon script for new HF config (#3448)

cebtenzzre committed 1 year ago

Verified 48edda30

build : use std::make_tuple() for compatibility with older GCC versions (#3488)

kenvix committed 1 year ago

Verified 45eba936

common : process escape sequences in reverse prompts (#3461)

staviq committed 1 year ago

Verified acec9eaa

CLBlast: Fix handling of on-device tensor data

shibe2 committed 1 year ago

e2583cbc

server : fix incorrect num_tokens_predicted (#3480)

jhen0409 committed 1 year ago

Verified e8b8d32e

swift : disable ACCELERATE_NEW_LAPACK (#3481)

jhen0409 committed 1 year ago

Verified 8f3a642e

ci : add swift build via xcodebuild (#3482)

jhen0409 committed 1 year ago

Verified 07453844

convert : fix Baichuan2 models by using vocab size in config.json (#3299)

KerfuffleV2 committed 1 year ago

Verified 019ba1dc

readme : add project status link

ggerganov committed 1 year ago

Verified beabc8cf

ggml : fix build after #3329

ggerganov committed 1 year ago

Verified 0d152b37

llm : add Refact model (#3329)

ds5t5 committed 1 year ago

Verified f8c90cdb

sync : ggml (conv 1d + 2d updates, UB fixes) (#3468)

ggerganov committed 1 year ago

Verified f93af024

finetune : readme fix typo (#3465)

iammerrick committed 1 year ago

Verified f72f8f22

Older