]> git.djapps.eu Git - pkg/ggml/sources/llama.cpp/commit
sampling : refactor init to use llama_sampling_params (#3696)
authorGeorgi Gerganov <redacted>
Fri, 20 Oct 2023 18:07:23 +0000 (21:07 +0300)
committerGitHub <redacted>
Fri, 20 Oct 2023 18:07:23 +0000 (21:07 +0300)
commitd1031cf49c3b958b915fd558e23453471c29ac33
tree14fa2bc6d54d5e27bd1e8bfd6fa4dbf894dbe6b9
parent8cf19d60dc93809db8e51fedc811595eed9134c5
sampling : refactor init to use llama_sampling_params (#3696)

* sampling : refactor init to use llama_sampling_params

* llama : combine repetition, frequency and presence penalties in 1 call

* examples : remove embd-input and gptneox-wip

* sampling : rename penalty params + reduce size of "prev" vector

* sampling : add llama_sampling_print helper

* sampling : hide prev behind API and apply #3661

ggml-ci
30 files changed:
Makefile
README.md
common/common.cpp
common/common.h
common/sampling.cpp
common/sampling.h
examples/CMakeLists.txt
examples/embd-input/.gitignore [deleted file]
examples/embd-input/CMakeLists.txt [deleted file]
examples/embd-input/README.md [deleted file]
examples/embd-input/embd-input-lib.cpp [deleted file]
examples/embd-input/embd-input-test.cpp [deleted file]
examples/embd-input/embd-input.h [deleted file]
examples/embd-input/embd_input.py [deleted file]
examples/embd-input/llava.py [deleted file]
examples/embd-input/minigpt4.py [deleted file]
examples/embd-input/panda_gpt.py [deleted file]
examples/gptneox-wip/cmpnct_gpt2bpe.hpp [deleted file]
examples/gptneox-wip/falcon-main.cpp [deleted file]
examples/gptneox-wip/gptneox-main.cpp [deleted file]
examples/infill/CMakeLists.txt
examples/infill/infill.cpp
examples/llava/llava-utils.h
examples/main/main.cpp
examples/parallel/parallel.cpp
examples/server/server.cpp
examples/speculative/speculative.cpp
llama.cpp
llama.h
tests/test-sampling.cpp