ggml-org/llama.cpp

Pull Requests Commits
server : return error on too large embedding input (#7389)
ggerganov committed 1 year ago
Verified e932094d
tests : fix --keep_split -> --keep-split (#7374)
ggerganov committed 1 year ago
Verified 2789baf4
Add provisions for windows support for BF16 code including CMake provision for enabling AVX512_BF16 (#7258)
Srihari-mcw committed 1 year ago
Verified 33c8d50a
llama : remove MPI backend (#7395)
slaren committed 1 year ago
Verified d359f309
quantize : fix --keep-split check (#7374)
fredlas committed 1 year ago
Verified 1ea2a003
Vulkan Embedding Fix (#7360)
0cc4m committed 1 year ago
Verified f030ec1f
ggml : fix another case of quants nans (#7387)
slaren committed 1 year ago
Verified e4e6f67b
ggml: implement quantized KV cache for FA (#7372)
JohannesGaessler committed 1 year ago
Verified 5ca49cbe
server: add test for token probs (#7347)
JohannesGaessler committed 1 year ago
Verified 1b01f06d
server: fix seed being reported back (#7382)
JohannesGaessler committed 1 year ago
Verified 41858392
Add StableLM2 pre-tokenizer (#7349)
aahouzi committed 1 year ago
Verified 6aade19e
cuda : clear error after buffer allocation failure (#7376)
slaren committed 1 year ago
Verified ab33f7a3
labeler.yml: Use settings from ggerganov/llama.cpp [no ci] (#7363)
mofosyne committed 1 year ago
Verified e23b974f
cmake : update android comments (#7341)
ggerganov committed 1 year ago
Verified 854d365a
Capture CUDA logging output (#7298)
fraxy-v committed 1 year ago
Verified f5bf7617
ci : re-enable sanitizer runs (#7358)
ggerganov committed 1 year ago
Verified 059031b8
android : use "ci-android" branch for CI (#7341)
ggerganov committed 1 year ago
Verified 511182ea
CUDA: deduplicate FlashAttention code (#7352)
JohannesGaessler committed 1 year ago
Verified 133d99c5
server: correct --threads documentation [no ci] (#7362)
JohannesGaessler committed 1 year ago
Verified cb42c294
cuda : add half2 __shfl_xor() for ROCm 5.5 (#7263)
Engininja2 committed 1 year ago
Verified d233b507
llama : add support for larger Granite Code Models (20B, 34B) (#7324)
sroecker committed 1 year ago
Verified 0f98acfa
perplexity : ndot progress and show stats with < 100 tasks (#7348)
strawberrymelonpanda committed 1 year ago
Verified ca57e0f3
Update and fix Vulkan soft_max and argsort implementations (#7237)
0cc4m committed 1 year ago
Verified c1b295ee
github-actions-labeler: initial commit (#7330)
mofosyne committed 1 year ago
Verified de731963
convert : fix set_vocab_sentencepiece (#6866)
ggerganov committed 1 year ago
Verified b49a13dd
ggml : fix quants nans when all the group weights are very close to zero (#7313)
slaren committed 1 year ago
Verified 05834841
cmake : fix typo in AMDGPU_TARGETS (#7356)
Engininja2 committed 1 year ago
Verified ef277de2
Unicode codepoint flags for custom regexs (#7245)
jaime-m-p committed 1 year ago
Verified b43272af
CUDA: faster large batch FA without tensor cores (#7314)
JohannesGaessler committed 1 year ago
Verified 0fc1e820
ROCm: use native CMake HIP support (#5966)
GZGavinZhao committed 1 year ago
Verified 82ca83db
Terms Of Use Privacy Policy Refund Policy Impressum