ngxson/llama.cpp

Pull Requests Commits

remove redundant check

ngxson committed 1 year ago

74e460d5

ngxson committed 1 year ago

9274a6bc

Apply suggestions from code review

ngxson committed 1 year ago

Verified a90e0642

move lora change task to queue

ngxson committed 1 year ago

1dbd16ab

ngxson committed 1 year ago

bf7df957

add slow test with llama 8b

ngxson committed 1 year ago

367f0ab1

Merge branch 'master' into xsn/lora_per_request

ngxson committed 1 year ago

d67fefb9

ggml : fixes for AVXVNNI instruction set with MSVC and Clang (#11027)

Srihari-mcw committed 1 year ago

Verified 0827b2c1

server : clean up built-in template detection (#11026)

ngxson committed 1 year ago

Verified 45095a61

server : add OAI compat for /v1/completions (#10974)

ngxson committed 1 year ago

Verified 5896c652

convert : fix Llama-3_1-Nemotron-51B rope settings (#11008)

ymcki committed 1 year ago

Verified bc7b1f86

common, examples, ggml : fix MSYS2 GCC compiler errors and warnings when building with LLAMA_CURL=ON and GGML_OPENCL=ON (#11013)

peter277 committed 1 year ago

Verified 6e1531ac

vulkan: optimize mul_mat for small values of N (#10991)

jeffbolznv committed 1 year ago

Verified 716bd6de

android : fix llama_batch free (#11014)

ag2s20150909 committed 1 year ago

Verified c250ecb3

vulkan: im2col and matmul optimizations for stable diffusion (#10942)

jeffbolznv committed 1 year ago

Verified a813badb

vulkan: Use push constant offset to handle misaligned descriptors (#10987)

jeffbolznv committed 1 year ago

Verified fdd21889

ngxson committed 1 year ago

076346db

server: added more docs for response_fields field (#10995)

isaac-mcfadyen committed 1 year ago

Verified f865ea14

server : fix token duplication when streaming with stop strings (#10997)

z80maniac committed 1 year ago

Verified 16cdce7b

move can_batch_with check

ngxson committed 1 year ago

b9b2b637

test: force disable cache prompt

ngxson committed 1 year ago

9947b077

lora per request

ngxson committed 1 year ago

9d84127f

slot.can_batch_with

ngxson committed 1 year ago

2ba6efc5

vulkan: multi-row k quants (#10846)

netrunnereve committed 1 year ago

Verified d79d8f39

examples, ggml : fix GCC compiler warnings (#10983)

peter277 committed 1 year ago

Verified d283d02b

server : add support for "encoding_format": "base64" to the */embeddings endpoints (#10967)

elk-cloner committed 1 year ago

Verified 9ba399df

ggml : more perfo with llamafile tinyblas on x86_64 (#10714)

Djip007 committed 1 year ago

Verified 2cd43f49

server: allow filtering llama server response fields (#10940)

nvrxq committed 1 year ago

Verified 09fe2e76

llama : the WPM vocabs use the CLS token as BOS (#10930)

ggerganov committed 1 year ago

Verified 30caac3a

ggml : use wstring for backend search paths (#10960)

slaren committed 1 year ago

Verified 60cfa728

Older