ngxson/llama.cpp

Pull Requests Commits

common : disable KV cache shifting automatically for unsupported models (#11053)

MollySophia committed 1 year ago

Verified 4b0c638b

metal : avoid uint (#11019)

ggerganov committed 1 year ago

Verified e7da954e

llama : refactor `src/llama.cpp` (#10902)

ggerganov committed 1 year ago

Verified f66f5829

server: bench: minor fixes (#10765)

phymbert committed 1 year ago

Verified 2f0ee84b

server : allow using LoRA adapters per-request (#10994)

ngxson committed 1 year ago

Verified 0da5d860

readme : add llama-swap to infrastructure section (#11032)

mostlygeek committed 1 year ago

Verified a45433ba

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

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

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

ggml : fix arm enabled features check (#10961)

slaren committed 1 year ago

Verified 3327bb0f

ggml : fix const usage in SSE path (#10962)

slaren committed 1 year ago

Verified 32d6ee63

server : fix missing model id in /model endpoint (#10957)

ngxson committed 1 year ago

Verified 14b699ec

server : add system_fingerprint to chat/completion (#10917)

ngxson committed 1 year ago

Verified 485dc012

rpc-server : add support for the SYCL backend (#10934)

rgerganov committed 1 year ago

Verified 86bf31cf

llama : support InfiniAI Megrez 3b (#10893)

dixyes committed 1 year ago

Verified b92a14a8

Newer Older