ggerganov/llama.cpp

Pull Requests Commits

convert-hf : support outtype templating in outfile name

compilade committed 1 year ago

e0af2df6

convert-hf : remove a semicolon because flake8 doesn't like it

compilade committed 1 year ago

d3d32a63

convert-hf : add --outtype auto-f16

compilade committed 1 year ago

58b515ca

convert-hf : get bit-exact same output as ./quantize

compilade committed 1 year ago

95930da3

convert-hf : add missing space after comma

compilade committed 1 year ago

3801db12

gguf-py : flake8 fixes

compilade committed 1 year ago

59f5a27f

convert-hf : support bfloat16 conversion

compilade committed 1 year ago

6f8d2800

cmake : fix typo (#7151)

cebtenzzre committed 1 year ago

Verified 4426e298

convert-hf : save memory with lazy evaluation (#7075)

compilade committed 1 year ago

Verified f98eb31c

Introduction of CUDA Graphs to LLama.cpp (#6766)

agray3 committed 1 year ago

Verified bc4bba36

JSON: [key] -> .at(key), assert() -> GGML_ASSERT (#7143)

JohannesGaessler committed 1 year ago

Verified c12452c7

Revert "llava : add support for moondream vision language model (#6899)"

ggerganov committed 1 year ago

Verified 9da243b3

server : add themes + favicon (#6848)

jboero committed 1 year ago

Verified bd1871fa

metal : use `vm_allocate` instead of `posix_memalign` on macOS (#7078)

giladgd committed 1 year ago

Verified 26458af1

main : add --conversation / -cnv flag (#7108)

Dawid Potocki committed 1 year ago

Verified 83330d8c

sgemm : AVX Q4_0 and Q8_0 (#6891)

netrunnereve committed 1 year ago

Verified 465263d0

server : add_special option for tokenize endpoint (#7059)

JohanAR committed 1 year ago

Verified 911b3900

convert.py : --vocab-only generates false but valid params (#7027)

20kdc committed 1 year ago

Verified ad211ede

llama : add BPE pre-tokenization for Qwen2 (#7114)

jklj077 committed 1 year ago

Verified 229ffff8

clean up json_value & server_log (#7142)

ngxson committed 1 year ago

Verified 1fd9c174

convert : add BPE pre-tokenization for DBRX (#7132)

dranger003 committed 1 year ago

Verified 4cd621c2

py : also print the normalizers

ggerganov committed 1 year ago

Verified 7e0b6a7b

compare-llama-bench.py: add missing basicConfig (#7138)

mofosyne committed 1 year ago

Verified acdce3cd

ggml : introduce bfloat16 support (#6412)

jart committed 1 year ago

Verified 38554160

metal : fix unused warning

ggerganov committed 1 year ago

Verified c0e6fbf8

Further tidy on Android instructions README.md (#7077)

Jeximo committed 1 year ago

Verified c780e753

Fixed save_imatrix to match old behaviour for MoE (#7099)

jukofyork committed 1 year ago

Verified 48b2f9c1

server: fix incorrectly reported token probabilities (#7125)

JohannesGaessler committed 1 year ago

Verified af0a5b61

Fix OLMo HF to GGUF conversion (#6910)

nopperl committed 1 year ago

Verified b6aa6702

server : update readme with undocumented options (#7013)

K-Mistele committed 1 year ago

Verified 260b7c65

Older