ngxson/llama.cpp

Pull Requests Commits
rm more unused
ngxson committed 1 year ago
0a7ef17c
rm unused kv
ngxson committed 1 year ago
1e4ce1d1
no more kv has_llava_projector
ngxson committed 1 year ago
5fa67231
clip : improve projector naming
ngxson committed 1 year ago
84a59220
llama : fix K-shift with quantized K and BLAS backend (#13113)
slaren committed 1 year ago
Verified 295354ea
Force FP32 compute in GLM4 FFN Down (#13101)
city96 committed 1 year ago
Verified 558a7647
clip : fix pixtral on some GPU backends (#13097)
ngxson committed 1 year ago
Verified edb18b6e
change the reorder tensor from init to execute OP (#13003)
NeoZhangJianyu committed 1 year ago
Verified 514c4560
rpc : do not wait for response when sending RPC_CMD_SET_TENSOR (#12943)
rgerganov committed 1 year ago
Verified 553a5c3a
clip : remove boi/eoi embeddings for GLM-edge model (#13081)
ngxson committed 1 year ago
Verified 13be08da
embeddings : fix batch sizes (#13076)
ggerganov committed 1 year ago
Verified 226251ed
ggml : fix trailing whitespaces (#0)
ggerganov committed 1 year ago
87616f06
sync : ggml
ggerganov committed 1 year ago
63b49114
ggml : Depthwise 2D convolution (ggml/1152)
Acly committed 1 year ago
c6e8cc28
CUDA: use switch statements in constexpr functions (#13095)
JohannesGaessler committed 1 year ago
Verified b10d8bfd
cmake : do not include ./src as public for libllama (#13062)
ggerganov committed 1 year ago
Verified 13b45488
clang-tidy : disable warning about missing math parenthesis (#13091)
ggerganov committed 1 year ago
Verified 572b3141
arg : add --no-mmproj-offload (#13093)
ngxson committed 1 year ago
Verified 7c727fbe
arg : clean up handling --mmproj with -hf (#13082)
ngxson committed 1 year ago
Verified 80982e81
metal : fix floating-point range of attention scores in FA kernels (#13090)
ggerganov committed 1 year ago
Verified 7604a7d6
vulkan: matmul gcn tuning (#13016)
netrunnereve committed 1 year ago
Verified b3b6d862
llama-mtmd-cli: Sigint rework in mtmd vision example (#13080)
pl752 committed 1 year ago
Verified 56304069
mtmd : Support Pixtral 12B (#13065)
ngxson committed 1 year ago
Verified ecda2ec4
convert : Append mult-eos,half-rope,bos to GLM4-0414 and Z (#13021)
piDack committed 1 year ago
Verified eb1776b1
rpc : add command line option for number of threads for the CPU backend (#13060)
rgerganov committed 1 year ago
Verified 2cca6c01
CUDA: noncont MMVQ + batched bs1 MUL_MAT_ID (#13014)
JohannesGaessler committed 1 year ago
Verified 658987cf
mtmd : support SmolVLM (version 1 and 2) (#13050)
ngxson committed 1 year ago
Verified dc39a5e7
security : add note about RPC and server functionality (#13061)
ggerganov committed 1 year ago
Verified ab47dec3
metal : add memory pool for temp allocs (#12850)
ggerganov committed 1 year ago
Verified 7b53389c
llava : update documentations (#13055)
ngxson committed 1 year ago
Verified 24345353
Terms Of Use Privacy Policy Refund Policy Impressum