Xuan-Son Nguyen
|
fe44d35574
tests : add test-jinja -py option for cross-checking (#18906)
|
пре 2 недеља |
Sigbjørn Skjæret
|
bbcdac0189
jinja : fix object item order (and properly implement dictsort) (#18904)
|
пре 2 недеља |
Sigbjørn Skjæret
|
d03c45c9c5
jinja : attribute support for join, map and sort (#18883)
|
пре 2 недеља |
Sigbjørn Skjæret
|
10c98cbdf6
jinja : add missing tojson filter for bool (#18900)
|
пре 2 недеља |
Sigbjørn Skjæret
|
420960ab92
jinja : fix lexing of float literals with sign (#18901)
|
пре 2 недеља |
Xuan-Son Nguyen
|
f55b033ae6
jinja: correct member access rule (#18905)
|
пре 2 недеља |
lhez
|
d1b4757ded
opencl: fix q6_K mv for m=1 (#18893)
|
пре 2 недеља |
Sigbjørn Skjæret
|
57c0beaed0
ci : add label for jinja changes (#18903)
|
пре 2 недеља |
Georgi Gerganov
|
2fbde785bc
kv-cache : optimize KQ mask construction (#18842)
|
пре 2 недеља |
Reese Levine
|
a89002f07b
ggml webgpu: support for backend sampling (#18880)
|
пре 2 недеља |
Thore Koritzius
|
388ce82241
ggml : extend ggml_pool_1d + metal (#16429)
|
пре 2 недеља |
hipudding
|
6ba6a3c76f
docs : update ops.md for CANN backend (#18654)
|
пре 2 недеља |
Perry Naseck
|
0802d4cfb3
ggml-blas: hide warnings from included BLAS headers (#18818)
|
пре 2 недеља |
Tarek Dakhran
|
c945aaaef2
mtmd : Fix ASR for LFM2.5-Audio-1.5B (#18876)
|
пре 2 недеља |
Xuan-Son Nguyen
|
c15395f73c
common : implement new jinja template engine (#18462)
|
пре 2 недеља |
Julius Tischbein
|
aa1dc3770a
Setting mmap and direct_io to false as default in llama-bench.cpp (#18841)
|
пре 2 недеља |
Raul Torres
|
4ea2eaac01
CANN: Remove unused `ggml_cann_get_device` function (#18625)
|
пре 2 недеља |
Chenguang Li
|
e20fa27a02
CANN: fix an issue where get_env was not fully renamed (#18796)
|
пре 2 недеља |
hipudding
|
baa4ba0aec
CANN: support gated linear attn (#18653)
|
пре 2 недеља |
shaofeiqi
|
785a710085
OpenCL: add SOLVE_TRI op support (#18846)
|
пре 2 недеља |
Georgi Gerganov
|
6e7fc8a146
cuda : print less debug logs when disabling cuda graphs (#18868)
|
пре 2 недеља |
Georgi Gerganov
|
be8e3d9515
context : do not reserve scheduler for warmups (#18867)
|
пре 2 недеља |
ddh0
|
13f1e4a9ca
llama : add adaptive-p sampler (#17927)
|
пре 2 недеља |
Xuan-Son Nguyen
|
a04c2b06a3
server: improve slots scheduling for n_cmpl (#18789)
|
пре 2 недеља |
Georgi Gerganov
|
39173bcacb
context : reserve new scheduler when graph topology changes (#18547)
|
пре 2 недеља |
Johannes Gäßler
|
5c662d21a3
CUDA: fix allignment on register spill for FA (#18815)
|
пре 2 недеља |
shalinib-ibm
|
8cc0ba957b
ggml-cpu: optimize ggml_vec_dot_bf16 for Power9 (#18837)
|
пре 2 недеља |
Xuan-Son Nguyen
|
a7e6ddb8bd
lora: make sure model keep track of associated adapters (#18490)
|
пре 2 недеља |
Sigbjørn Skjæret
|
2a13180100
model-loader : support bool array sliding window pattern (#18850)
|
пре 2 недеља |
Adrien Gallouët
|
ec997b4f2b
tests : download models only when running ctest (#18843)
|
пре 2 недеља |