ggerganov/llama.cpp

Pull Requests Commits

Kawrakow committed 1 year ago

d34472c1

Faster perplexity computation (#2786)

ikawrakow committed 1 year ago

Verified d046dcee

llama : add llama_beam_search() (#2267)

mattpulver committed 1 year ago

Verified c82742ac

convert.py : Get rope scale from HuggingFace models (#2772)

pnb committed 1 year ago

Verified 28b2c996

llama-bench : add model sizes (#2771)

slaren committed 1 year ago

Verified 154725c5

convert.py : export rope freq_base when converting CodeLlama from an HF model (#2773)

slaren committed 1 year ago

Verified 12e2e33a

server : display token probabilities in the UI (#2489)

jhen0409 committed 1 year ago

Verified 29674ab4

ci : pip install gguf in editable mode (#2782)

ggerganov committed 1 year ago

Verified 5439a0ab

gguf : export objects to user code (#2780)

monatis committed 1 year ago

Verified 8194cd87

ROCm Port (#1087)

SlyEcho committed 1 year ago

Verified 6bbc598a

cuda : add RoPE kernel for mode == 2 (NeoX) (#2760)

ggerganov committed 1 year ago

Verified 3f460a2b

gguf : make gguf pip-installable

monatis committed 1 year ago

Verified 87e3733f

ggml-alloc : enlarge size of parse_seq (#2776)

lshzh-ww committed 1 year ago

Verified b91ad7f4

Added `enum` to `llama_token_get_type` return type (#2774)

MarcusDunn committed 1 year ago

Verified 2e5f70a2

convert.py : try to determine n_ctx automatically for CodeLlama (#2770)

slaren committed 1 year ago

Verified d0f77b13

gguf : add rope_freq_base parameter for CodeLlama (#2769)

slaren committed 1 year ago

Verified 0d3094f0

falcon : write file type

ggerganov committed 1 year ago

Verified 01f22246

metal : bug-fix when enable ggml-alloc (#2757)

lshzh-ww committed 1 year ago

Verified 38b16dfc

convert : auto-determine model name based on dir + scripts update

ggerganov committed 1 year ago

Verified 8f8c28e8

Fix for main example getting stuck when -n -2 and --interactive (#2767)

KerfuffleV2 committed 1 year ago

Verified 7694adda

fix convert.py for codellama, add llama 34B to the list of recognized models (#2768)

slaren committed 1 year ago

Verified fea95c68

Tag release with build number (#2732)

DannyDaemonic committed 1 year ago

Verified ef955fbd

metal : add Q8_0 support (#2763)

ggerganov committed 1 year ago

Verified d67777c2

llama : escape all U+2581 in a string (#2750)

ggerganov committed 1 year ago

Verified c3e53b42

llama : fix grammar sometimes generating null char (#2756)

ejones committed 1 year ago

Verified 6e91a1b0

readme : fix link

ggerganov committed 1 year ago

Verified 44d5462b

minor : fix trailing whitespace

ggerganov committed 1 year ago

Verified c7868b07

readme : update hot topics

ggerganov committed 1 year ago

Verified 79da24b5

llm : add Falcon support (#2717)

ggerganov committed 1 year ago

Verified cf658adc

minor : fix trailing whitespace

ggerganov committed 1 year ago

Verified a192860c

Older