chat: faster KV shift, continue generating, fix stop sequences (#2781)

* Don't stop generating at end of context
* Use llama_kv_cache ops to shift context
* Fix and improve reverse prompt detection
* Replace prompt recalc callback with a flag to disallow context shift
This commit is contained in:
Jared Van Bortel
2024-08-07 11:25:24 -04:00
committed by GitHub
parent 90de2d32f8
commit be66ec8ab5
16 changed files with 285 additions and 230 deletions

View File

@@ -33,7 +33,7 @@ set(LLMODEL_VERSION_PATCH 0)
set(LLMODEL_VERSION "${LLMODEL_VERSION_MAJOR}.${LLMODEL_VERSION_MINOR}.${LLMODEL_VERSION_PATCH}")
project(llmodel VERSION ${LLMODEL_VERSION} LANGUAGES CXX C)
set(CMAKE_CXX_STANDARD 20)
set(CMAKE_CXX_STANDARD 23)
set(CMAKE_CXX_STANDARD_REQUIRED ON)
set(CMAKE_LIBRARY_OUTPUT_DIRECTORY ${CMAKE_RUNTIME_OUTPUT_DIRECTORY})
set(BUILD_SHARED_LIBS ON)