]> git.djapps.eu Git - pkg/ggml/sources/llama.cpp/commit
Introduce C-style API (#370)
authorGeorgi Gerganov <redacted>
Wed, 22 Mar 2023 05:32:36 +0000 (07:32 +0200)
committerGitHub <redacted>
Wed, 22 Mar 2023 05:32:36 +0000 (07:32 +0200)
commitf5a77a629bd0f37ae1696747633ab42a5530ec15
treeb3d147dd228ce67661ed497a6dc61b444a38e0f9
parentda0e9fe90ccf6e73597eb19dd0cfc0a28363fb3b
Introduce C-style API (#370)

* Major refactoring - introduce C-style API

* Clean up

* Add <cassert>

* Add <iterator>

* Add <algorithm> ....

* Fix timing reporting and accumulation

* Measure eval time only for single-token calls

* Change llama_tokenize return meaning
14 files changed:
CMakeLists.txt
Makefile
convert-pth-to-ggml.py
ggml.c
ggml.h
llama.cpp [new file with mode: 0644]
llama.h [new file with mode: 0644]
main.cpp
models/ggml-vocab.bin
quantize.cpp
tests/CMakeLists.txt
tests/test-tokenizer-0.cpp
utils.cpp
utils.h