ggerganov/llama.cpp

Pull Requests Commits
vulkan: Implement grouped query attention in the coopmat2 FA shader (#12559)
jeffbolznv committed 1 year ago
Verified be0a0f8c
Vulkan: Fix mmq int dot float cache size (#12722)
0cc4m committed 1 year ago
Verified 92e3006b
model : print tensor size during load (#12711)
ggerganov committed 1 year ago
Verified 833e2b74
llama : add option to override model tensor buffers (#11397)
slaren committed 1 year ago
Verified e0e912f4
llama : refactor kv cache guard (#12695)
ggerganov committed 1 year ago
Verified a10b36c9
vocab : BailingMoE : change possessive quantifiers to greedy (#12677)
CISC committed 1 year ago
Verified 83a88bd6
common : remove json.hpp from common.cpp (#12697)
ngxson committed 1 year ago
Verified 42eb248f
[CANN] get_rows and dup optimization (#12671)
noemotiovon committed 1 year ago
Verified 9bacd6b3
common : refactor downloading system, handle mmproj with -hf option (#12694)
ngxson committed 1 year ago
Verified 267c1399
opencl : fix memory allocation size (#12649)
sparkleholic committed 1 year ago
Verified f423981a
llama : use LLM_KV_GENERAL_FILE_TYPE instead of gguf_find_key (#12672)
jklincn committed 1 year ago
Verified e39e727e
convert : BailingMoE : fix qkv split when head_dim is 0 (#12687)
CISC committed 1 year ago
Verified 5936a616
metal : use F32 prec in FA kernels (#12688)
ggerganov committed 1 year ago
Verified 3fd072a5
Fix clang warning in gguf_check_reserved_keys (#12686)
yeahdongcn committed 1 year ago
Verified a6f32f0b
vulkan: fix build when glslc doesn't support coopmat (#12683)
wbruna committed 1 year ago
Verified 2bb3597e
SYCL: Rename oneMKL to oneMath (#12192)
Rbiessy committed 1 year ago
Verified 82939705
SYCL: switch to SYCL namespace (#12674)
qnixsynapse committed 1 year ago
Verified 8bbf2608
convert : BailingMoE : avoid setting rope_dim to 0 (#12678)
CISC committed 1 year ago
Verified 35782aee
vocab : add special infill tokens for CodeLlama (#11850)
danbev committed 1 year ago
Verified c80a7759
ggml : faster ssm scan (#10558)
A3shTnT committed 1 year ago
Verified 250d7953
convert : Qwerky : use lora_rank_tokenshift and lora_rank_decay if present (#12667)
CISC committed 1 year ago
Verified 403fbacb
Vulkan: Add DP4A MMQ and Q8_1 quantization shader (#12135)
0cc4m committed 1 year ago
Verified a8a1f335
cmake : fix whitespace (#0)
ggerganov committed 1 year ago
1790e731
sync : ggml
ggerganov committed 1 year ago
0114a32d
cmake: improve Vulkan cooperative matrix support checks (whisper/2966)
sandrohanea committed 1 year ago
a7724480
llava : proper description fix (#12668)
CISC committed 1 year ago
Verified 1a859490
SYCL: Remove misleading ggml_sycl_op_flatten function (#12387)
qnixsynapse committed 1 year ago
Verified 6c02a032
llava : fix clip loading GGUFs with missing description (#12660)
CISC committed 1 year ago
Verified f52d59d7
tts : remove printfs (#12640)
marcoStocchi committed 1 year ago
Verified 52de2e59
llama : support BailingMoE (Ling) (#12634)
CISC committed 1 year ago
Verified 2c3f8b85
Terms Of Use Privacy Policy Refund Policy Impressum