ngxson/llama.cpp

Pull Requests Commits

ngxson committed 1 year ago

b9ef1bc7

ngxson committed 1 year ago

d6e882be

add server bench script docker

ngxson committed 1 year ago

537cdd3a

Fix a spelling mistake (#9001)

Septa2112 committed 1 year ago

Verified 2589292c

py : fix requirements check '==' -> '~=' (#8982)

ggerganov committed 1 year ago

Verified d3ae0ee8

server : handle models with missing EOS token (#8997)

ggerganov committed 1 year ago

Verified 5ef07e25

gguf-py : Numpy dequantization for most types (#8939)

compilade committed 1 year ago

Verified 4134999e

flake.lock: Update (#8979)

ggerganov committed 1 year ago

Verified 8cd1bcfd

update guide (#8909)

arthw committed 1 year ago

Verified a21c6fd4

llama : check all graph nodes when searching for result_embd_pooled (#8956)

fairydreaming committed 1 year ago

Verified 33309f66

Optimize Vulkan backend for better CPU performance and less GPU synchronization overhead. (#8943)

mtavenrath committed 1 year ago

Verified 7c5bfd57

metal : fix uninitialized abort_callback (#8968)

slaren committed 1 year ago

Verified 6e02327e

llama : default n_swa for phi-3 (#8931)

ngxson committed 1 year ago

Verified 7eb23840

Add support for encoder-only T5 models (#8900)

fairydreaming committed 1 year ago

Verified 7c3f55c1

gguf-py : fix double call to add_architecture() (#8952)

tarilabs committed 1 year ago

Verified 911b437f

Merge commit from fork

ggerganov committed 1 year ago

Verified b72942fa

llama : add support for lora adapters in T5 model (#8938)

fairydreaming committed 1 year ago

Verified 6afd1a99

make : fix llava obj file race (#8946)

ggerganov committed 1 year ago

Verified 272e3bd9

llama : better replace_all (cont) (#8926)

ggerganov committed 1 year ago

Verified 45a55b91

llava : support MiniCPM-V-2.5 (#7599)

tc-mb committed 1 year ago

Verified 3071c0a5

ggerganov committed 1 year ago

Verified 4305b57c

whisper : use vulkan as gpu backend when available (whisper/2302)

mstephenson6 committed 1 year ago

Verified 70c0ea35

embedding : add --pooling option to README.md [no ci] (#8934)

danbev committed 1 year ago

Verified 5b2c04f4

llama : fix typo in llama_tensor_get_type comment [no ci] (#8937)

danbev committed 1 year ago

Verified 6f6496bb

server : add one level list nesting for embeddings (#8936)

gelim committed 1 year ago

Verified daef3ab2

llama : reduce useless copies when saving session (#8916)

compilade committed 1 year ago

Verified 345a686d

gguf-py : simplify support for quant types (#8838)

compilade committed 1 year ago

Verified 3a14e003

scripts : sync cann files (#0)

ggerganov committed 1 year ago

Verified afd27f01

scripts : fix sync filenames (#0)

ggerganov committed 1 year ago

Verified 366d486c

ggerganov committed 1 year ago

Verified e44a561a

Older