ggml-org/llama.cpp

Pull Requests Commits
SYCL: disable faulty fp16 exp kernel (#14395)
qnixsynapse committed 1 year ago
Verified f47c1d71
ggml : fix unmerged GGML_FPxx_TO_FPxx refactoring (#14443)
CISC committed 1 year ago
Verified a5d1fb62
ggml : implement REGLU/GEGLU/SWIGLU ops (#14158)
CISC committed 1 year ago
Verified a0535ffa
vulkan: Add fusion support for RMS_NORM+MUL (#14366)
jeffbolznv committed 1 year ago
Verified bd9c981d
CUDA: add bf16 and f32 support to cublas_mul_mat_batched (#14361)
am17an committed 1 year ago
Verified 27208bf6
vulkan: handle noncontig in the final case of ggml_vk_get_cpy_pipeline (#14378)
jeffbolznv committed 1 year ago
Verified 63a7bb3c
vulkan: lock accesses of pinned_memory vector (#14333)
jeffbolznv committed 1 year ago
Verified 00d5282c
model : add support for ERNIE 4.5 0.3B model (#14408)
ownia committed 1 year ago
Verified 566c16fc
fix async_mode bug (#14432)
bachelor-dou committed 1 year ago
Verified b25e9277
ci : fix windows build and release (#14431)
CISC committed 1 year ago
Verified 6609507a
vulkan: Fix GGML_VULKAN_SHADER_DEBUG_INFO (#14427)
jeffbolznv committed 1 year ago
Verified ceb1bf5a
graph : make llm_graph_context destructor virtual (#14410)
ggerganov committed 1 year ago
Verified 72babea5
recurrent : call balloc split_reset() in init_batch() (#14414)
ggerganov committed 1 year ago
Verified 43678060
ggml : add ggml_set_rows (#14274)
rgerganov committed 1 year ago
Verified 8d94219a
convert : fix broken sentencepiece vocab (#14416)
CISC committed 1 year ago
Verified f667f1e6
model : gemma3n text-only (#14400)
ngxson committed 1 year ago
Verified 8846aace
cmake: regen vulkan shaders when shaders-gen sources change (#14398)
mtmcp committed 1 year ago
Verified a01047b0
llama : return mistral-v7-tekken as default template only (#14390)
CISC committed 1 year ago
Verified b2534622
metal : add special-case mat-vec mul for ne00 == 4 (#14385)
ggerganov committed 1 year ago
Verified e8215dbb
metal : batch rows copy in a single threadgroup (#14384)
ggerganov committed 1 year ago
Verified 5783ae43
docs: update s390x documentation + add faq (#14389)
taronaeo committed 1 year ago
Verified bf5bcd0b
musa: enable fp16 mma (all) and cublas on qy2 (#13842)
yeahdongcn committed 1 year ago
Verified 716301d1
ggml-cpu: enable IBM NNPA Vector Intrinsics (#14317)
taronaeo committed 1 year ago
Verified 60ef23d6
ggml : do not output unprintable characters on GGUF load failure (#14381)
CISC committed 1 year ago
Verified b193d530
sycl: GGML_SYCL_DISABLE_OPT on by default for all Intel Devices (#13973)
ShanoToni committed 1 year ago
Verified 2bf9d539
opencl: ref count `ggml_backend_opencl_context` and refactor profiling (#14254)
lhez committed 1 year ago
Verified 73e53dc8
batch : fix check for empty sequences in memory (#14364)
ggerganov committed 1 year ago
Verified 62af4642
cmake : use LLAMA_BUILD_NUMBER when defining LLAMA_INSTALL_VERSION (#14362)
mbaudier committed 1 year ago
Verified c148cf19
server : move no API key doc to /health (#14352)
pnb committed 1 year ago
Verified 1b809cee
main : honor --verbose-prompt on interactive prompts (#14350)
CISC committed 1 year ago
Verified abf24104
Terms Of Use Privacy Policy Refund Policy Impressum