ngxson/llama.cpp

Pull Requests Commits
Merge branch 'master' into xsn/webui_pako
ngxson committed 1 year ago
59107110
webui: Allow pasting file from clipboard (#13526)
luca020400 committed 1 year ago
Verified d486dd3e
docs: Update link to ggml-org in multimodal.md (#13513)
ddpasa committed 1 year ago
Verified 21ca987f
use fflate instead of pako
ngxson committed 1 year ago
9383d11e
simpler code
ngxson committed 1 year ago
1413bbb9
scripts : fix compare-llama-bench.py show parameter (#13514)
CISC committed 1 year ago
Verified be1d4a13
webui : use pako for more deterministic gzip compress
ngxson committed 1 year ago
63788265
vulkan: workaround FA compile failures on macos (#13517)
jeffbolznv committed 1 year ago
Verified ab3971f2
quantize : improve tensor-type pattern matching (#13033)
EAddario committed 1 year ago
Verified e5c834f7
clip : clip.h become private API (⚠️ breaking change) (#13510)
ngxson committed 1 year ago
Verified 71bdbdb5
metal : use FA-vec kernel up to batch size 20 (#13496)
ggerganov committed 1 year ago
Verified f0995d28
metal : optimize multi-sequence FA vec kernel (#13493)
ggerganov committed 1 year ago
Verified c252e0c4
ggml-cpu: Update KleidiAI to v1.6 and fix include directives (#13509)
eddnjjn committed 1 year ago
Verified 4f711afe
batched-bench : fix pp batch contents (#13492)
ggerganov committed 1 year ago
Verified b89d605a
mtmd : remove libllava, remove clip-quantize-cli (⚠️ breaking change) (#13460)
ngxson committed 1 year ago
Verified b4726345
scripts : support arbitrary input file formats in compare-llama-bench.py (#13455)
CISC committed 1 year ago
Verified bf793711
model : Granite MoE shared (#13269)
gabe-l-hart committed 1 year ago
Verified d590cd4c
sync : ggml
ggerganov committed 1 year ago
1e2809bc
llama-bench : add defrag-thold, check for invalid ranges (#13487)
slaren committed 1 year ago
Verified cf0a43bb
opencl: remove unnecessary assert for `add` (#13257)
lhez committed 1 year ago
Verified f0d46ef1
clip : cap max image size 1024 for qwen vl model (#13478)
ngxson committed 1 year ago
Verified de4c07f9
llama/ggml: add LLM training support (#10544)
JohannesGaessler committed 1 year ago
Verified 10d2af0e
context : fix state io for memory-less contexts (#13470)
ggerganov committed 1 year ago
Verified 064cc596
server : allow content to be null in oaicompat_completion_params_parse (#13477)
anudit committed 1 year ago
Verified 91159ee9
llama-bench : accept ranges for integer parameters (#13410)
slaren committed 1 year ago
Verified 22cdab34
ggml-cpu: Integrate fp32=bf16xbf16 SME KleidiAI kernel (#13053)
eddnjjn committed 1 year ago
Verified a71a4075
CUDA: fix misaligned synchronization in FA (#13469)
JohannesGaessler committed 1 year ago
Verified 95e18884
ggml : add mrope kernel for metal (#13457)
ngxson committed 1 year ago
Verified df849192
enable dpcpp nightly builds with libraries (#13406)
AD2605 committed 1 year ago
Verified 14492144
mtmd : Use RMS norm for InternVL 3 38B and 78B mmproj (#13459)
city96 committed 1 year ago
Verified c1040239
Terms Of Use Privacy Policy Refund Policy Impressum