ochafik/llama.cpp

Pull Requests Commits
sampling : make top_n_sigma no-op at <=0 or a single candidate (#13345)
DocShotgun committed 1 year ago
Verified ffc72720
sampling : don't consider -infinity values in top_n_sigma (#13344)
oobabooga committed 1 year ago
Verified 91a86a6f
cmake : remove arm64 msvc presets (#13342)
slaren committed 1 year ago
Verified f4ed10b6
SYCL: Disable reorder optimize by default and stop setting tensor extras when optimize is disabled (#13254)
qnixsynapse committed 1 year ago
Verified 1e333d5b
llama : fix build_ffn without gate (#13336)
ngxson committed 1 year ago
Verified 2f54e348
CUDA: fix bad asserts for partial offload (#13337)
JohannesGaessler committed 1 year ago
Verified 2356fb1d
convert : qwen2/3moe : set yarn metadata if present (#13331)
CISC committed 1 year ago
Verified 764b8562
CUDA: fix --split-mode row for MMQ (#13323)
JohannesGaessler committed 1 year ago
Verified 15a28ec8
gguf-py : avoid requiring pyside6 for other scripts (#13036)
compilade committed 1 year ago
Verified a7366faa
CUDA: fix logic for clearing padding with -ngl 0 (#13320)
JohannesGaessler committed 1 year ago
Verified 90703650
sampling : Integrate Top-nσ into main sampling chain (and add it to the server) (#13264)
oobabooga committed 1 year ago
Verified 233461f8
server : Webui - change setText command from parent window to also send the message. (#13309)
igardev committed 1 year ago
Verified b34c8591
mtmd : rename llava directory to mtmd (#13311)
ngxson committed 1 year ago
Verified 9b61acf0
clip : fix confused naming ffn_up and ffn_down (#13290)
ngxson committed 1 year ago
Verified 5215b91e
convert : bailingmoe : set yarn metadata if present (#13312)
CISC committed 1 year ago
Verified ae803bfc
SYCL: Disable mul_mat kernels for noncontiguous tensor b (#13308)
qnixsynapse committed 1 year ago
Verified 66645a52
mtmd : add C public API (#13184)
ngxson committed 1 year ago
Verified 27aa2595
rpc : use backend registry, support dl backends (#13304)
slaren committed 1 year ago
Verified 9fdfcdae
ggml : activate s390x simd for Q3_K (#13301)
taronaeo committed 1 year ago
Verified 6eb7d25c
llava/mtmd : fixes to fully support dl backends (#13303)
slaren committed 1 year ago
Verified 86bd60d3
llama : build windows releases with dl backends (#13220)
slaren committed 1 year ago
Verified 9f2da587
CUDA: fix race condition in MMQ stream-k fixup (#13299)
JohannesGaessler committed 1 year ago
Verified 93c4e239
CUDA: fix race condition in MMQ ids_dst (#13294)
JohannesGaessler committed 1 year ago
Verified 8afbd968
vulkan: Additional type support for unary, binary, and copy (#13266)
jeffbolznv committed 1 year ago
Verified 8ae5ebcf
imatrix: fix oob writes if src1 is not contiguous (#13286)
JohannesGaessler committed 1 year ago
Verified 3e959f09
clip : revert the change of BOI/EOI token for GLM-edge (⚠️ breaking change) (#13259)
ngxson committed 1 year ago
Verified 36667c8e
llama : Llama-3_1-Nemotron-Ultra-253B-v1 support (#12843)
ymcki committed 1 year ago
Verified 3bf785f3
llama : move end-user examples to tools directory (#13249)
slaren committed 1 year ago
Verified 1d36b367
sync : ggml (#13268)
ggerganov committed 1 year ago
Verified b3444392
context : fix reorder logic (#13267)
ggerganov committed 1 year ago
Verified a75cb30d
Terms Of Use Privacy Policy Refund Policy Impressum