ggerganov/llama.cpp

Pull Requests Commits

convert.py : advanced option (#2753)

KerfuffleV2 committed 2 years ago

Verified 730d9c68

llama : use Unicode Escape Sequence to replace encoded characters (#2814)

drasticactions committed 2 years ago

Verified c7d92e6d

flake.nix : add rocm support and cleanup (#2808)

Tungsten842 committed 2 years ago

Verified 61d1a289

llama : move #includes out of _GNU_SOURCE conditional (#2817)

cebtenzzre committed 2 years ago

Verified 741ca7dd

main : fix bug (penalize_nl=false doesn't work) + suppress warning on mingw (#1528)

tom7 committed 2 years ago

Verified 72f895c9

llama : use std::abs in llama_sample_tail_free (#2800)

cebtenzzre committed 2 years ago

Verified 50526f37

k-quants : remove unnecessary tensor shape restrictions (#2811)

ggerganov committed 2 years ago

Verified 04f4b1eb

Better perplexity for 2- and 3-bit quantization for LLaMA-v2-70B (#2807)

ikawrakow committed 2 years ago

Verified 75923754

Fix HellaSwag (#2805)

ikawrakow committed 2 years ago

Verified 771551a7

flake : build llama.cpp on Intel with nix (#2795)

signalpillar committed 2 years ago

Verified f305bad1

Handle null rope scaling value (#2793)

pnb committed 2 years ago

Verified a2ca4e9d

Fix spm whitespaces (#2806)

klosax committed 2 years ago

Verified 2ba83c86

examples : skip unnecessary external lib in server README.md how-to (#2804)

longregen committed 2 years ago

Verified bae5c5f6

llama : fix struct decl (#2790)

MarcusDunn committed 2 years ago

Verified 232caf3c

Faster perplexity computation (#2786)

ikawrakow committed 2 years ago

Verified d046dcee

llama : add llama_beam_search() (#2267)

mattpulver committed 2 years ago

Verified c82742ac

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

pnb committed 2 years ago

Verified 28b2c996

llama-bench : add model sizes (#2771)

slaren committed 2 years ago

Verified 154725c5

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

slaren committed 2 years ago

Verified 12e2e33a

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

jhen0409 committed 2 years ago

Verified 29674ab4

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

ggerganov committed 2 years ago

Verified 5439a0ab

gguf : export objects to user code (#2780)

monatis committed 2 years ago

Verified 8194cd87

ROCm Port (#1087)

SlyEcho committed 2 years ago

Verified 6bbc598a

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

ggerganov committed 2 years ago

Verified 3f460a2b

gguf : make gguf pip-installable

monatis committed 2 years ago

Verified 87e3733f

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

lshzh-ww committed 2 years ago

Verified b91ad7f4

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

MarcusDunn committed 2 years ago

Verified 2e5f70a2

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

slaren committed 2 years ago

Verified d0f77b13

gguf : add rope_freq_base parameter for CodeLlama (#2769)

slaren committed 2 years ago

Verified 0d3094f0

falcon : write file type

ggerganov committed 2 years ago

Verified 01f22246

Newer Older