ggerganov/llama.cpp

Pull Requests Commits

Mtmd: add a way to select device for vision encoder (#14236)

stduhpf committed 53 days ago

Verified c8ade300

cuda : implement bf16 cpy ops and enable bf16 cont (#14763)

CISC committed 53 days ago

Verified e28c0b80

opencl: remove unreachable `return` (#14806)

lhez committed 54 days ago

Verified 8e6f8bc8

server : allow setting `--reverse-prompt` arg (#14799)

MollySophia committed 54 days ago

Verified adef8178

cuda: remove linking to cublasLt (#14790)

yeahdongcn committed 54 days ago

Verified 48b86c4f

opencl: fix `im2col` when `KW!=KH` (#14803)

CISC committed 54 days ago

Verified 38d3af1b

opencl: add conv2d kernel (#14403)

rmatif committed 54 days ago

Verified 6c9ee3b1

sycl: Fix im2col (#14797)

Rbiessy committed 54 days ago

Verified cd465d82

kleidiai: add support for get_rows (#14676)

chaxu01 committed 54 days ago

Verified 92204260

docs : fix backends table in README.md (#14796)

rgerganov committed 54 days ago

Verified 2ba1333b

vulkan/cuda: Fix im2col when KW!=KH (#14789)

jeffbolznv committed 54 days ago

Verified c2e058f1

llama : fix `--reverse-prompt` crashing issue (#14794)

MollySophia committed 55 days ago

Verified c82d48ec

server : add parse_special option to /tokenize endpoint (#14783)

IsaacDynamo committed 55 days ago

Verified b4efd77f

docs : fix link for tools/perplexity in README.md (#14780)

am17an committed 55 days ago

Verified 2be60cbc

Documentation: Further revisions to the Vulkan section in build.md (#14785)

rspOverflow committed 55 days ago

Verified b526ad26

Clang-format: local files first + fix BinPacking (#14779)

am17an committed 55 days ago

Verified 938b7857

Contrib: add 0cc4m as codeowner for Vulkan backend (#14775)

0cc4m committed 56 days ago

Verified 36c15324

ggml: adds CONV_2D op and direct GEMM Vulkan implementation (#14316)

etasnadi committed 56 days ago

Verified a979ca22

imatrix : use a single count for dense 3d tensors

compilade committed 56 days ago

73439beb

imatrix : use GGUF to store importance matrices (#9400)

compilade committed 56 days ago

Verified 90083283

vulkan: Add logging for bf16 features to ggml_vk_print_gpu_info (#13274) (#14707)

Peter0x44 committed 56 days ago

Verified d4b91ea7

Vulkan: Fix fprintf format-security warning (#14770)

0cc4m committed 56 days ago

Verified 83f58724

Documentation: Update build.md's Vulkan section (#14736)

rspOverflow committed 56 days ago

Verified f0d4d176

ggerganov committed 57 days ago

b1723091

metal : fuse add, mul + add tests (#14596)

ggerganov committed 57 days ago

Verified bf9087f5

graph : fix graph reuse reset of params (#14760)

ggerganov committed 57 days ago

Verified 9fb1042c

parallel : add option for different RNG seeds (#14757)

ggerganov committed 57 days ago

Verified 2adf8d83

cuda : Fix Gemma3n not executed as CUDA_GRAPH on NVGPUs (#14741)

ORippler committed 57 days ago

Verified 021cc28b

graph : avoid huge warm-up graphs for MoE models (#14753)

ggerganov committed 57 days ago

Verified d498af3d

model : fix build after merge conflict (#14754)

ggerganov committed 58 days ago

Verified eacdeb5b

Newer Older