ggml-org/llama.cpp

Pull Requests Commits

examples : add compare-mlx

ggerganov committed 6 days ago

Verified 5b01d857

lookup, lookahead: fix crash when n_ctx not specified (#18729)

pestopoppa committed 6 days ago

Verified 14883391

ngram-mod : fix build [no ci] (#19216)

ggerganov committed 7 days ago

Verified 49277958

opencl: add optimized q8_0 mm kernel for adreno (#18871)

shaofeiqi committed 7 days ago

Verified 971facc3

ggerganov committed 7 days ago

d9a2a4bc

cuda : fix compile warnings (whisper/0)

ggerganov committed 7 days ago

dfd6106c

server : wrap around the "id_slot" parameter (#19207)

ggerganov committed 7 days ago

Verified bbada8bf

Correctly fetch q8_1 quantize pipeline in test as needed by 8a3519b (#19194)

sredman committed 7 days ago

Verified 13f3ebfa

spec : add ngram-mod (#19164)

ggerganov committed 7 days ago

Verified dabaa2e7

jinja : add unordered_map include to value.h [no ci] (#19205)

mseri committed 7 days ago

Verified 2e916f99

memory : clarify comments for r_l and s_l tensors [no ci] (#19203)

danbev committed 7 days ago

Verified f3bc9889

tests : add GQA=20 FA test (#19095)

ggerganov committed 7 days ago

Verified c3b87ceb

convert : add missing return statement for GraniteMoeModel (#19202)

danbev committed 7 days ago

Verified 05625031

memory : remove unused tmp_buf (#19199)

danbev committed 7 days ago

Verified 83bcdf72

docs: Add LlamaLib to UI projects (#19181)

amakropoulos committed 7 days ago

Verified b316895f

add tensor type checking as part of cuda graph properties (#19186)

bssrdf committed 7 days ago

Verified ecbf01d4

sycl: implement GGML_UNARY_OP_SOFTPLUS (#19114)

s8322 committed 7 days ago

Verified 1025fd2c

sycl: implement GGML_OP_TRI (#19089)

RachelMantel committed 7 days ago

Verified c7358ddf

Fix typos in SYCL documentation (#19162)

DDXDB committed 7 days ago

Verified d284baf1

ggml-webgpu: improve flastAttention performance by software pipelining (#19151)

ArberSephirotheca committed 7 days ago

Verified bd90fc74

hexagon: enable offloading to Hexagon on Windows on Snapdragon (#19150)

tboinovski1 committed 7 days ago

Verified ce38a4db

cuda : fix nkvo, offload and cuda graph node properties matching (#19165)

ggerganov committed 8 days ago

Verified 4fdbc1e4

chat : add parsing for solar-open-100b (#18540)

aldehir committed 8 days ago

Verified 7b7ae857

webui: Update Svelte to fix effect_update_depth_exceeded errors (#19144)

amarshall committed 8 days ago

Verified 84b0a983

jinja : do not pass empty tools and add some none filters (#19176)

CISC committed 8 days ago

Verified b45ef270

HIP: add mmf for CDNA (#18896)

zhang-hui-yulo committed 8 days ago

Verified f3dd7b8e

arg : add -kvu to llama-batched-bench (#19172)

ggerganov committed 8 days ago

Verified eed25bc6

ggml-zendnn : resolve ZenDNN backend cross-module symbol dependency (#19159)

z-vishal committed 8 days ago

Verified b33df266

CUDA: refactor topk-moe to enable more models (GLM 4.7, Nemotron etc.) (#19126)

am17an committed 8 days ago

Verified 3bcc9909

sycl: fix norm kernels: l2_norm, group_norm, rms_norm by remove assert to support more cases (#19154)

arthw committed 8 days ago

Verified d4964a7c

Older