ngxson/llama.cpp

Pull Requests Commits

add llama_lora_adapter_clear

ngxson committed 2 years ago

b41cbd08

server : fix URL.parse in the UI (#8646)

0x4139 committed 2 years ago

Verified b841d074

sycl : Add support for non-release DPC++ & oneMKL (#8644)

joeatodd committed 2 years ago

Verified 64cf50a0

llama : move vocab, grammar and sampling into separate files (#8508)

ggerganov committed 2 years ago

Verified 938943cd

Vulkan IQ4_NL Support (#8613)

0cc4m committed 2 years ago

Verified 751fcfc6

Allow all RDNA2 archs to use sdot4 intrinsic (#8629)

jeroen-mostert committed 2 years ago

Verified 46e47417

contrib : clarify PR squashing + module names (#8630)

ggerganov committed 2 years ago

Verified e7e6487b

[SYCL] fix scratch size of softmax (#8642)

luoyu-intel committed 2 years ago

Verified 063d99ad

llama : fix codeshell support (#8599)

hankeke303 committed 2 years ago

Verified 081fe431

llama : add support for SmolLm pre-tokenizer (#8609)

Stillerman committed 2 years ago

Verified d94c6e0c

*.py: Stylistic adjustments for python (#8233)

jpodivin committed 2 years ago

Verified 566daa5a

llama : allow overrides for tokenizer flags (#8614)

ggerganov committed 2 years ago

Verified 6f11a83e

tests : re-enable tokenizer tests (#8611)

ggerganov committed 2 years ago

Verified e093dd23

llama : add Mistral Nemo inference support (#8604)

iamlemec committed 2 years ago

Verified 50e05353

server : update doc to clarify n_keep when there is bos token (#8619)

kaetemi committed 2 years ago

Verified 62815449

ggml: fix compile error for RISC-V (#8623)

zqb-all committed 2 years ago

Verified 04bab6b7

examples: fix android example cannot be generated continuously (#8621)

devojony committed 2 years ago

Verified b7c11d36

flake.lock: Update (#8610)

ggerganov committed 2 years ago

Verified 45f2c19c

examples : Rewrite pydantic_models_to_grammar_examples.py (#8493)

maruel committed 2 years ago

Verified 22f281aa

gguf-py : fix some metadata name extraction edge cases (#8591)

compilade committed 2 years ago

Verified 328884f4

convert_hf : fix Gemma v1 conversion (#8597)

compilade committed 2 years ago

Verified c69c6303

CUDA: MMQ code deduplication + iquant support (#8495)

JohannesGaessler committed 2 years ago

Verified 69c487f4

gguf : handle null name during init (#8587)

ggerganov committed 2 years ago

Verified 07283b1a

llama : add support for Tekken pre-tokenizer (#8579)

m18coppola committed 2 years ago

Verified 94036222

llama.swiftui: fix end of generation bug (#8268)

ho2103 committed 2 years ago

Verified 69b9945b

gguf_dump.py: fix markddown kv array print (#8588)

mofosyne committed 2 years ago

Verified c3776cac

ggml : fix quant dot product with odd number of blocks (#8549)

slaren committed 2 years ago

Verified 87e397d0

convert-*.py: remove add_name from ChatGLMModel class (#8590)

mofosyne committed 2 years ago

Verified 57b1d4f9

llama : bump max layers from 256 to 512 (#8530)

ggerganov committed 2 years ago

Verified d1975455

readme : fix server badge

ggerganov committed 2 years ago

Verified be0cfb41

Older