ggml-org/llama.cpp

Pull Requests Commits

test-model-random : fix seq_id buffer overflow

compilade committed 66 days ago

7634d14d

test-model-random : add shared prompt test variant

compilade committed 66 days ago

a17c4f7d

test-model-random : avoid testing too many sequences for now

compilade committed 66 days ago

4e58ca46

Merge branch 'master' into compilade/test-model-random

compilade committed 66 days ago

18d20551

test-model-random : configurable model n_ctx, and smaller seq lengths

compilade committed 66 days ago

362cf542

vulkan: optimize flash attention split_k_reduce (#14554)

jeffbolznv committed 66 days ago

Verified 6efcd659

model : fix hunyuan moe chat template (#14584)

stevenkuang-tencent committed 66 days ago

Verified 699f4392

model : add SmolLM3 (#14581)

ngxson committed 66 days ago

Verified 08382869

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

compilade committed 66 days ago

Verified bb4f7a9e

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

jeffbolznv committed 66 days ago

Verified b8eeb874

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

oluwandabira committed 67 days ago

Verified 17a1f0d2

model : add hunyuan moe (#14425)

ngxson committed 67 days ago

Verified 8f22dc0a

vulkan: increase timeout for CI (#14574)

jeffbolznv committed 67 days ago

Verified 53903ae6

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

ggerganov committed 67 days ago

Verified 4d0dcd4a

CUDA: add bilinear interpolation for upscale (#14563)

am17an committed 67 days ago

Verified 75c91de6

memory : fix broken batch splits for recurrent cache

compilade committed 67 days ago

6b38c7a0

test-model-random : add Mamba2

compilade committed 67 days ago

985cda6c

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

yeahdongcn committed 67 days ago

Verified 68155c66

Merge branch 'master' into compilade/test-model-random

compilade committed 67 days ago

48a5eba5

llama : fix incorrect minicpm3 v_states shape (#14571)

CISC committed 67 days ago

Verified e1a70590

llama : remove ggml_cont where possible (#14568)

CISC committed 67 days ago

Verified 12f55c30

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

am17an committed 67 days ago

Verified b9c3eefd

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

netrunnereve committed 68 days ago

Verified 6491d6e4

vulkan: fix rms_norm+mul fusion (#14545)

jeffbolznv committed 69 days ago

Verified e592be15

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

jeffbolznv committed 70 days ago

Verified a0374a67

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

CISC committed 70 days ago

Verified ddef9952

opencl: add GELU_ERF (#14476)

CISC committed 70 days ago

Verified 66816881

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

ggerganov committed 70 days ago

Verified bac8bed2

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

yeahdongcn committed 70 days ago

Verified b81510a7

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

ggerganov committed 70 days ago

Verified ef797db3

Older