ggerganov/llama.cpp

Pull Requests Commits

speculative : experimenting with Qwen2.5

ggerganov committed 294 days ago

Verified 5e6dad93

speculative : fix out-of-bounds access

ggerganov committed 294 days ago

Verified 33bdee66

vulkan: Optimize binary ops (#10270)

jeffbolznv committed 295 days ago

Verified af148c93

vulkan: Use macros to make the mat mul pipeline creation more concise (#10259)

jeffbolznv committed 295 days ago

Verified 66798e42

llama : propagate the results of `graph_compute` (#9525)

Xarbirus committed 295 days ago

Verified fb4a0ec0

ggerganov committed 295 days ago

Verified 5ea926da

docs : update bindings list (#10261)

xuegao-tzx committed 295 days ago

Verified 1ee9eea0

server : add missing docs (#10269)

z80maniac committed 295 days ago

Verified ff7fb670

server : fix incorrect res in validate_model_chat_template (#10272)

jhen0409 committed 295 days ago

Verified 0e712a5a

metadata: Detailed Dataset Authorship Metadata (#8875)

mofosyne committed 295 days ago

Verified a0ec17b3

sycl : Fixes to broken builds and test-backend-ops (#10257)

Alcpz committed 295 days ago

Verified 2e82ffa4

vulkan: Optimize contiguous copies (#10254)

jeffbolznv committed 296 days ago

Verified 80dd7ff2

vulkan: Throttle the number of shader compiles during the build step. (#10222)

jeffbolznv committed 297 days ago

Verified 54ef9cfc

metal : more precise Q*K in FA vec kernel (#10247)

ggerganov committed 298 days ago

Verified b0cefea5

server : enable KV cache defrag by default (#10233)

ggerganov committed 298 days ago

Verified b141e5f6

flake.lock: Update (#10243)

ggerganov committed 298 days ago

Verified 4b3a9212

server : (web UI) Add back sampler settings (#10239)

MaggotHATE committed 298 days ago

Verified 505f3327

vulkan: Fix newly added tests for permuted mul_mat and 1D im2col (#10226)

jeffbolznv committed 298 days ago

Verified 160687b3

metal : reorder write loop in mul mat kernel + style (#10231)

ggerganov committed 299 days ago

Verified 6423c65a

metal : fix build and some more comments (#10229)

ggerganov committed 299 days ago

Verified 39a334a9

metal : fix F32 accumulation in FA vec kernel (#10232)

ggerganov committed 299 days ago

Verified bb38cdd8

llama : fix Qwen model type strings

ggerganov committed 299 days ago

Verified f018acba

metal : hide debug messages from normal log

ggerganov committed 299 days ago

Verified 46323fa9

ggml: fix zero division in ‘dne’ calculation in CUDA COUNT_EQUAL operator when ‘ne’ is small (#10213)

SongXiaoXi committed 300 days ago

Verified 5b359bb1

ggml : optimize llamafile cpu matrix multiplication for ppc64le (#10156)

amritahs-ibm committed 300 days ago

Verified e8921349

scripts : fix pattern and get n_tokens in one go (#10221)

lhpqaq committed 300 days ago

Verified 8fc393f2

metal : opt-in compile flag for BF16 (#10218)

ggerganov committed 300 days ago

Verified ec450d3b

metal : improve clarity (minor) (#10171)

ggerganov committed 300 days ago

Verified 695ad752

metal : optimize FA kernels (#10171)

ggerganov committed 300 days ago

Verified 841f27ab

swift : exclude ggml-metal-embed.metal (#10211)

jhen0409 committed 300 days ago

Verified d05b3127

Older