ggml-org/llama.cpp

Pull Requests Commits
Allow all RDNA2 archs to use sdot4 intrinsic (#8629)
jeroen-mostert committed 1 year ago
Verified 46e47417
contrib : clarify PR squashing + module names (#8630)
ggerganov committed 1 year ago
Verified e7e6487b
[SYCL] fix scratch size of softmax (#8642)
luoyu-intel committed 1 year ago
Verified 063d99ad
llama : fix codeshell support (#8599)
hankeke303 committed 1 year ago
Verified 081fe431
llama : add support for SmolLm pre-tokenizer (#8609)
Stillerman committed 1 year ago
Verified d94c6e0c
*.py: Stylistic adjustments for python (#8233)
jpodivin committed 1 year ago
Verified 566daa5a
llama : allow overrides for tokenizer flags (#8614)
ggerganov committed 1 year ago
Verified 6f11a83e
tests : re-enable tokenizer tests (#8611)
ggerganov committed 1 year ago
Verified e093dd23
llama : add Mistral Nemo inference support (#8604)
iamlemec committed 1 year ago
Verified 50e05353
server : update doc to clarify n_keep when there is bos token (#8619)
kaetemi committed 1 year ago
Verified 62815449
ggml: fix compile error for RISC-V (#8623)
zqb-all committed 1 year ago
Verified 04bab6b7
examples: fix android example cannot be generated continuously (#8621)
devojony committed 1 year ago
Verified b7c11d36
flake.lock: Update (#8610)
ggerganov committed 1 year ago
Verified 45f2c19c
examples : Rewrite pydantic_models_to_grammar_examples.py (#8493)
maruel committed 1 year ago
Verified 22f281aa
gguf-py : fix some metadata name extraction edge cases (#8591)
compilade committed 1 year ago
Verified 328884f4
convert_hf : fix Gemma v1 conversion (#8597)
compilade committed 1 year ago
Verified c69c6303
CUDA: MMQ code deduplication + iquant support (#8495)
JohannesGaessler committed 1 year ago
Verified 69c487f4
gguf : handle null name during init (#8587)
ggerganov committed 1 year ago
Verified 07283b1a
llama : add support for Tekken pre-tokenizer (#8579)
m18coppola committed 1 year ago
Verified 94036222
llama.swiftui: fix end of generation bug (#8268)
ho2103 committed 1 year ago
Verified 69b9945b
gguf_dump.py: fix markddown kv array print (#8588)
mofosyne committed 1 year ago
Verified c3776cac
ggml : fix quant dot product with odd number of blocks (#8549)
slaren committed 1 year ago
Verified 87e397d0
convert-*.py: remove add_name from ChatGLMModel class (#8590)
mofosyne committed 1 year ago
Verified 57b1d4f9
llama : bump max layers from 256 to 512 (#8530)
ggerganov committed 1 year ago
Verified d1975455
readme : fix server badge
ggerganov committed 1 year ago
Verified be0cfb41
ggml : add friendlier error message to fopen errors (#8575)
HanClinto committed 1 year ago
Verified b57eb9ca
fix: typo of chatglm4 chat tmpl (#8586)
thxCode committed 1 year ago
Verified f299aa98
convert-*.py: add general.name kv override (#8571)
mofosyne committed 1 year ago
Verified 3d0e4367
CUDA: fix partial offloading for ne0 % 256 != 0 (#8572)
JohannesGaessler committed 1 year ago
Verified a15ef8f8
cmake : install all ggml public headers (#8480)
65a committed 1 year ago
Verified 705b7ecf
Terms Of Use Privacy Policy Refund Policy Impressum