ggml-org/llama.cpp

Pull Requests Commits

cont : cache shift support

ggerganov committed 18 days ago

Verified ff76c673

cont : rotate caches separately + support non-power-of-2 head sizes

ggerganov committed 18 days ago

Verified 7711b3a3

cont : rotate V more + refactor

ggerganov committed 18 days ago

Verified 832e3263

llama : rotate activations for better quantization

ggerganov committed 18 days ago

Verified e5aa067d

common : filter out imatrix when finding models (#21023)

angt committed 18 days ago

Verified 9900b29c

fix(ggml): correct RISC-V ISA string canonical ordering for RVV in CMake (#20888)

ihb2032 committed 19 days ago

Verified dc8d14c5

common : make LLAMA_CACHE the one cache for everything (#21009)

angt committed 19 days ago

Verified 93dfbc12

common : fix split model migration (#21019)

angt committed 19 days ago

Verified 3cba8bba

ggml-cuda: Add NVFP4 dp4a kernel (#20644)

michaelw9999 committed 19 days ago

Verified 112c7815

imatrix : fix crash when using --show-statistics with zero counts (#19532)

ssam18 committed 19 days ago

Verified 0fac87b1

CUDA & CPU: support F32 kernel type for `CONV_TRANSPOSE_2D` (#17094)

AgainstEntropy committed 19 days ago

Verified 0a524f24

common : do not delete old files from the old cache when updating (#21000)

angt committed 19 days ago

Verified c0159f9c

mtmd: Add DeepSeekOCR Support (#17400)

sfallah committed 19 days ago

Verified a970515b

common : fix verbosity setup (#20989)

angt committed 19 days ago

Verified 056b50c3

common : fix gguf selection in common_list_cached_models (#20996)

angt committed 19 days ago

Verified f2c72b8f

ci : fix parsing of vgpr counts in hip-quality-check (#20987)

IMbackK committed 19 days ago

Verified ec54ac13

model: codefuse-ai/F2LLM-v2 support

sfallah committed 19 days ago

Verified 80322ebd

model : allow causal_attn and pooling_type on all architectures (#20973)

Bing-su committed 19 days ago

Verified 44c51e52

snapdragon: add missing features to WoS scripts to achieve parity with ADB scripts (#20884)

aparmp-quic committed 19 days ago

Verified 1922f87c

Use docker in build-android.yml (#20928)

shreyajn committed 19 days ago

Verified 345de3cd

llama-bench: print `-n-cpu-moe` when offloaded layers > 1 (#20984)

am17an committed 20 days ago

Verified 9c600bcd

ci: Allow ninja to be used during unit test (#20742)

rillomas committed 20 days ago

Verified b2704f90

ci : disable self-hosted mac jobs (#20985)

ggerganov committed 20 days ago

Verified 3fab96cd

jinja: fix macro with kwargs (#20960)

ngxson committed 20 days ago

Verified 914eb5ff

gguf-split : clarify operation of gguf-split (#19749)

alosslessdev committed 20 days ago

Verified 8fc17493

llama: fix llama-model-saver (#20503)

JohannesGaessler committed 20 days ago

Verified 36dafba5

webui: Fix editing assistant message without branching (#20944)

allozaur committed 20 days ago

Verified 69e0ecef

Add SLEEPING status to the WebUI model selector (#20949)

ServeurpersoCom committed 20 days ago

Verified 062cca58

android : fix-pointer-dangling (#20974)

yikechayedan committed 20 days ago

Verified 406f4e3f

sycl : fix wrong variable check by assert (#20903)

arthw committed 20 days ago

Verified 53dc8b59

Older