ngxson/llama.cpp

Pull Requests Commits

cuda : enable CUDA graphs for MMID 1 <= BS <= 4 (#19645)

ggerganov committed 94 days ago

Verified ad8207af

model-conversion : make printing of config values optional (#19681)

danbev committed 94 days ago

Verified 667b6942

ci : bump komac version (#19682)

CISC committed 94 days ago

Verified e48349a4

build : link ws2_32 as PUBLIC on Windows (#19666)

angt committed 94 days ago

Verified ae46a61e

build : cleanup library linking logic (#19665)

angt committed 94 days ago

Verified 65cede7c

convert : add JoyAI-LLM-Flash (#19651)

dranger003 committed 94 days ago

Verified 05fa625e

perplexity: add proper batching (#19661)

AesSedai committed 94 days ago

Verified d6129011

common : inline functions (#18639)

Nekotekina committed 94 days ago

Verified cceb1b4e

ggml : make `ggml_is_view` as API (#19539)

foldl committed 94 days ago

Verified d23a5599

model: Add support for Tiny Aya Models (#19611)

saurabhdash2512 committed 94 days ago

Verified 5f28c53d

build : rework llama_option_depr to handle LLAMA_CURL (#19658)

angt committed 94 days ago

Verified 44084941

Adjust workaround for ROCWMMA_FATTN/GFX9 to only newer ROCm veresions (#19591)

superm1 committed 94 days ago

Verified 2ba9adc0

models : deduplicate delta-net graphs for Qwen family (#19597)

ggerganov committed 94 days ago

Verified cc45f2ad

graph : fix KQ mask, lora, cvec reuse checks (#19644)

ggerganov committed 95 days ago

Verified d5dfc330

ggml: aarch64: Implement SVE in Gemm q4_k 8x8 q8_k Kernel (#19132)

abhijain1204fujitsu committed 95 days ago

Verified 267ba5a1

ggerganov committed 95 days ago

ff4affb4

ggml : bump version to 0.9.7 (ggml/1425)

ggerganov committed 95 days ago

55d58599

ggml : bump version to 0.9.6 (ggml/1423)

ggerganov committed 95 days ago

1a8c700b

cuda: optimize iq2xxs/iq2xs/iq3xxs dequantization (#19624)

dfriehs committed 95 days ago

Verified 27b93cbd

docs: update s390x build docs (#19643)

taronaeo committed 95 days ago

Verified 6e67fd21

build : remove LLAMA_HTTPLIB option (#19623)

angt committed 95 days ago

Verified 9e118b97

cmake : check if KleidiAI API has been fetched (#19640)

danbev committed 95 days ago

Verified 57088276

context : fix output reorder with backend sampling (#19638)

ggerganov committed 95 days ago

Verified 341bc7d2

ggml : avoid UB in gemm ukernel (#19642)

ggerganov committed 95 days ago

Verified 08e6d914

ggml-cpu: optimize ggml_vec_dot_bf16 for s390x (#19399)

taronaeo committed 96 days ago

Verified 184c694f

ggml-cpu: FA add GEMM microkernel (#19422)

am17an committed 96 days ago

Verified 684b3610

cmake : fix KleidiAI install target failure with EXCLUDE_FROM_ALL (#19581)

ssam18 committed 96 days ago

Verified 3a00c985

convert : ensure all models handle new experts count (#19621)

CISC committed 96 days ago

Verified 079feab9

mtmd : Add Nemotron Nano 12B v2 VL support (#19547)

anavp-nvidia committed 96 days ago

Verified 01d8eaa2

models : optimize qwen3next graph (#19375)

ggerganov committed 97 days ago

Verified 1725e316

Newer Older