ggerganov/llama.cpp

Pull Requests Commits
llama : enable flash attn automatically when supported
slaren committed 1 year ago
afc4a7de
ggml : fix memory leaks when loading invalid gguf files (#10094)
slaren committed 1 year ago
Verified b9e02e81
readme : more lora detail in main example readme (#10064)
richdougherty committed 1 year ago
Verified 6763f713
convert : more detailed convert lora usage docs (#10065)
richdougherty committed 1 year ago
Verified 79a2bc04
ggml : add Q4_0_8_8 RISC-V GEMV and GEMM kernels (#10029)
xctan committed 1 year ago
Verified fc83a9e5
llama : refactor model loader with backend registry (#10026)
slaren committed 1 year ago
Verified c5b0f4b5
ggml: Add POOL2D OP for GPU acceleration to the Vulkan backend in the MobileVLM model. (#9763)
cyzero-kim committed 1 year ago
Verified 8f275a7c
llama : remove Tail-Free sampling (#10071)
ggerganov committed 1 year ago
Verified 8d8ff715
llama : Add IBM granite template (#10013)
arch-btw committed 1 year ago
Verified 61715d5c
flake.lock: Update (#10063)
ggerganov committed 1 year ago
Verified 07028f9d
musa: workaround for Guilty Lockup in cleaning src0 (#10042)
yeahdongcn committed 1 year ago
Verified 524afeec
server : don't overfill the batch during infill (#10018)
ggerganov committed 1 year ago
Verified 8125e6cb
llama : switch KQ multiplication to F32 precision by default (#10015)
ggerganov committed 1 year ago
Verified 8841ce3f
sync : ggml
ggerganov committed 1 year ago
Verified cc2983d3
increase cuda_cpy block size (ggml/996)
bssrdf committed 1 year ago
Verified 8c60a8a4
scripts : fix amx sync [no ci]
ggerganov committed 1 year ago
Verified 9e4a2563
metal : support permuted matrix multiplicaions (#10033)
ggerganov committed 1 year ago
Verified 66875035
llama : add DRY sampler (#9702)
wwoodsTM committed 1 year ago
Verified ff252ea4
llama: string_split fix (#10022)
Xarbirus committed 1 year ago
Verified d80fb71f
llamafile : extend sgemm.cpp support for Q5_0 models (#10010)
Srihari-mcw committed 1 year ago
Verified 2f8bd2b9
server : check that the prompt fits in the slot's context (#10030)
ggerganov committed 1 year ago
Verified bc5ba007
server : refactor slot input data, move tokenizer to HTTP thread (#10023)
ngxson committed 1 year ago
Verified 958367bf
ci : fix cmake flags for SYCL
ggerganov committed 1 year ago
Verified 40f25557
CUDA: fix insufficient buffer clearing for MMQ (#10032)
JohannesGaessler committed 1 year ago
Verified 167a5156
CUDA: fix MMQ for non-contiguous src0, add tests (#10021)
JohannesGaessler committed 1 year ago
Verified c39665f5
server : samplers accept the prompt correctly (#10019)
wwoodsTM committed 1 year ago
Verified 0a1c750c
sync : ggml
ggerganov committed 1 year ago
Verified 190a37d7
llama.vim : bump generation time limit to 3s [no ci]
ggerganov committed 1 year ago
Verified 2d3aba9e
CUDA: fix 1D im2col, add tests (ggml/993)
JohannesGaessler committed 1 year ago
80273a30
ggml : remove redundant set of contexts used field (ggml/978)
danbev committed 1 year ago
c19af0ac
Terms Of Use Privacy Policy Refund Policy Impressum