ggml-org/llama.cpp

Pull Requests Commits
server : add parameter -tb N, --threads-batch N (#3584) (#3768)
cebtenzzre committed 2 years ago
Verified ad939626
server : do not block system prompt update (#3767)
ggerganov committed 2 years ago
Verified 1717521c
sync : ggml (conv ops + cuda MSVC fixes) (#3765)
ggerganov committed 2 years ago
Verified b2f7e04b
cmake : add missed dependencies (#3763)
kingsidelee committed 2 years ago
Verified abd21fc9
cuda : add batched cuBLAS GEMM for faster attention (#3749)
ggerganov committed 2 years ago
Verified 2b4ea35e
Add more tokenizer tests (#3742)
Galunid committed 2 years ago
Verified daab3d7f
metal : handle ggml_scale for n%4 != 0 (close #3754)
ggerganov committed 2 years ago
Verified 469c9add
Revert "make : add optional CUDA_NATIVE_ARCH (#2482)"
ggerganov committed 2 years ago
Verified e3932593
issues : separate bug and enhancement template + no default title (#3748)
monatis committed 2 years ago
Verified 9d029564
Update special token handling in conversion scripts for gpt2 derived tokenizers (#3746)
Galunid committed 2 years ago
Verified 69a67350
llama : remove token functions with `context` args in favor of `model` (#3720)
MarcusDunn committed 2 years ago
Verified 5be6c803
Fix baichuan convert script not detecing model (#3739)
Galunid committed 2 years ago
Verified 6336701c
make : add optional CUDA_NATIVE_ARCH (#2482)
awhill19 committed 2 years ago
Verified 96981f37
server : parallel decoding and multimodal (#3677)
ggerganov committed 2 years ago
Verified 438c2ca8
Add test for MPT tokenization (#3728)
goerch committed 2 years ago
Verified 9e70cc03
readme : remove unsupported node.js library (#3703)
ianscrivener committed 2 years ago
Verified 5a42a5f8
llama : validate special token ids are in range when loading GGUF model (#3635)
KerfuffleV2 committed 2 years ago
Verified a5e7dbd6
main : escape prompt for cfg_negative_prompt and consecutive inputs in main with interactive (#3623)
vvhg1 committed 2 years ago
Verified d3956aea
batched : add len CLI argument
ggerganov committed 2 years ago
Verified 22c69a27
CLBlast: Add outer loops over src0 for broadcasting in mulmat
shibe2 committed 2 years ago
465219b9
sampling : refactor init to use llama_sampling_params (#3696)
ggerganov committed 2 years ago
Verified d1031cf4
gguf : support big endian platform (#3552)
chenqiny committed 2 years ago
Verified 8cf19d60
server : fix uninitialized sampling context (close #3685)
ggerganov committed 2 years ago
Verified a0edf73b
ggml : fix rope + llama minor optimizations (#3560)
GermanAizek committed 2 years ago
Verified f439e506
convert : restore compat with old Falcon models (#3680)
cebtenzzre committed 2 years ago
Verified e78f3ef2
multimodal : add BakLLaVA conversion support (#3682)
monatis committed 2 years ago
Verified f3b25e40
llava : avoid segfault in case of non-existent mmproj file (#3674)
monatis committed 2 years ago
Verified 60abea97
readme : update hot topics
ggerganov committed 2 years ago
Verified 004797f6
speculative : bug fixes
ggerganov committed 2 years ago
Verified 4e82b2ea
speculative : add tree-based sampling example (#3624)
ggerganov committed 2 years ago
Verified 0e89203b
Terms Of Use Privacy Policy Refund Policy Impressum