ggerganov/llama.cpp

Pull Requests Commits

cmake : use ggml-metal.metal from source dir to build default.metallib

cebtenzzre committed 264 days ago

b979fc97

cuda : fix defrag with quantized KV (#9319)

slaren committed 264 days ago

Verified 4db04784

llama-bench : fix NUL terminators in CPU name (#9313)

slaren committed 265 days ago

Verified bdf314f3

ggml : AVX2 support for Q4_0_8_8 (#8713)

Srihari-mcw committed 265 days ago

Verified 581c3051

[SYCL] Fix DMMV dequantization (#9279)

OuadiElfarouki committed 265 days ago

Verified 5910ea94

Fix broken links in docker.md (#9306)

carlory committed 265 days ago

Verified c8671ae2

rpc : make RPC servers come first in the device list (#9296)

rgerganov committed 265 days ago

Verified 82e3b03c

readme : rename result_format to response_format (#9300)

iscy committed 265 days ago

Verified 9379d3cc

flake.lock: Update (#9261)

ggerganov committed 266 days ago

Verified 7605ae7d

llama-bench : add JSONL (NDJSON) output mode (#9288)

akx committed 266 days ago

Verified 8962422b

readme : refactor API section + remove old hot topics

ggerganov committed 266 days ago

Verified b69a480a

server : test script : add timeout for all requests (#9282)

ngxson committed 267 days ago

Verified 48baa61e

src: make tail invalid when kv cell is intersection for mamba (#9249)

kylo5aby committed 267 days ago

Verified f1485161

docker : fix missing binaries in full-cuda image (#9278)

slaren committed 267 days ago

Verified 048de848

ggml : add pthread includes on FreeBSD (#9258)

yurivict committed 267 days ago

Verified f771d064

server : refactor multitask handling (#9274)

ngxson committed 267 days ago

Verified 6e7d133a

llama-cli : remove duplicated log message (#9275)

nbcsm committed 267 days ago

Verified b60074f1

build(nix): Package gguf-py (#5664)

ditsuke committed 267 days ago

Verified 9c1ba557

llama : minor style

ggerganov committed 267 days ago

Verified c6d4cb46

llama : support RWKV v6 models (#8980)

MollySophia committed 268 days ago

Verified 8f1d81a0

nix: fix CUDA build - replace deprecated autoAddOpenGLRunpathHook

enolan committed 269 days ago

a47667cf

sgemm : improved Q4_0 and Q8_0 performance via 4xN and Mx4 gemm (#8908)

Srihari-mcw committed 269 days ago

Verified ea5d7478

llama : fix typo in xcda_array_view comment [no ci] (#9132)

danbev committed 269 days ago

Verified 49271efb

llama : fix llama_split_mode enum values in main_gpu document (#9057)

kou committed 270 days ago

Verified 0ab30f8d

Correct typo run_llama2.sh > run-llama2.sh (#9149)

shou692199 committed 270 days ago

Verified cddae488

llava : the function "clip" should be int (#9237)

tc-mb committed 270 days ago

Verified 7ea8d80d

Threadpool: take 2 (#8672)

fmz committed 271 days ago

Verified 42c76d13

server : fix crash when error handler dumps invalid utf-8 json (#9195)

kaetemi committed 271 days ago

9f7d4bcf

flake.lock: Update (#9162)

ggerganov committed 271 days ago

Verified 1d1ccce6

docker : build images only once (#9225)

slaren committed 272 days ago

Verified 9fe94cca

Older