ggerganov/llama.cpp

Pull Requests Commits

cuda : support Falcon-H1 state size for SSM_SCAN

compilade committed 92 days ago

11807528

ggml : prevent integer overflow in gguf tensor size calculation (#14595)

Yuuoniy committed 92 days ago

Verified 26a48ad6

model : add skt/A.X-4.0 model vocabulary (#14589)

Bing-su committed 92 days ago

Verified ffd59e7d

llama : remove unintended whitespace (#14592)

CISC committed 92 days ago

Verified 10555459

model : add support for Falcon-H1 family (#14534)

ibrahimkhadraoui committed 92 days ago

Verified 04655063

convert : fix smollm3 jinja template (#14586)

ngxson committed 92 days ago

Verified 20b7bf8a

vulkan: optimize flash attention split_k_reduce (#14554)

jeffbolznv committed 93 days ago

Verified 6efcd659

model : fix hunyuan moe chat template (#14584)

stevenkuang-tencent committed 93 days ago

Verified 699f4392

model : add SmolLM3 (#14581)

ngxson committed 93 days ago

Verified 08382869

memory : fix broken batch splits for recurrent cache (#14575)

compilade committed 93 days ago

Verified bb4f7a9e

vulkan : fix rope with partial rotation and non-cont src (#14582)

jeffbolznv committed 93 days ago

Verified b8eeb874

server: Add ability to mount server at prefix (#14544)

oluwandabira committed 93 days ago

Verified 17a1f0d2

model : add hunyuan moe (#14425)

ngxson committed 93 days ago

Verified 8f22dc0a

vulkan: increase timeout for CI (#14574)

jeffbolznv committed 93 days ago

Verified 53903ae6

cuda : fix rope with partial rotation and non-cont src (#14580)

ggerganov committed 93 days ago

Verified 4d0dcd4a

CUDA: add bilinear interpolation for upscale (#14563)

am17an committed 93 days ago

Verified 75c91de6

musa: fix build warnings (unused variable) (#14561)

yeahdongcn committed 94 days ago

Verified 68155c66

llama : fix incorrect minicpm3 v_states shape (#14571)

CISC committed 94 days ago

Verified e1a70590

llama : remove ggml_cont where possible (#14568)

CISC committed 94 days ago

Verified 12f55c30

CUDA: add bf16 and i32 to getrows (#14529)

am17an committed 94 days ago

Verified b9c3eefd

vulkan: increase LOAD_VEC_A to 8 (IQ1/IQ2) or 4 (IQ3) (#14485)

netrunnereve committed 95 days ago

Verified 6491d6e4

vulkan: fix rms_norm+mul fusion (#14545)

jeffbolznv committed 95 days ago

Verified e592be15

vulkan: Handle updated FA dim2/3 definition (#14518)

jeffbolznv committed 96 days ago

Verified a0374a67

server : fix assistant prefilling when content is an array (#14360)

CISC committed 96 days ago

Verified ddef9952

opencl: add GELU_ERF (#14476)

CISC committed 96 days ago

Verified 66816881

eval-callback : check for empty input (#14539)

ggerganov committed 96 days ago

Verified bac8bed2

test-backend-ops: add support for specifying output format (#14368)

yeahdongcn committed 96 days ago

Verified b81510a7

metal : disable fast math in all quantize kernels (#14528)

ggerganov committed 97 days ago

Verified ef797db3

batch : add optional for sequential equal split (#14511)

ggerganov committed 97 days ago

Verified 67d1ef23

graph : prepare for 4D mask (#14515)

ggerganov committed 97 days ago

Verified 7b50f7c0

Older