diff --git a/.github/FUNDING.yml b/.github/FUNDING.yml new file mode 100644 index 00000000..97476fa7 --- /dev/null +++ b/.github/FUNDING.yml @@ -0,0 +1 @@ +buy_me_a_coffee: fractalego \ No newline at end of file diff --git a/documentation/source/installation.rst b/documentation/source/installation.rst index b6a794a5..e8ff1cb3 100644 --- a/documentation/source/installation.rst +++ b/documentation/source/installation.rst @@ -31,7 +31,7 @@ The second command starts the audio interface as well as a web server on port 80 Please see the examples in the following chapters. -LLM side (needs a GPU) +LLM side (needs a GPU to be efficient) ---------------------- The second part (LLM side) is a model server for the speech-to-text model, the LLM, the embedding system, and the text-to-speech model. In order to quickly run the LLM side, you can use the following installation commands: @@ -41,7 +41,7 @@ In order to quickly run the LLM side, you can use the following installation com $ pip install wafl-llm $ wafl-llm start - which will use the default models and start the server on port 8080. +which will use the default models and start the server on port 8080. The interface side has a `config.json` file that needs to be filled with the IP address of the LLM side. The default is localhost. diff --git a/todo.txt b/todo.txt index 8e2dd6a9..9acd21db 100644 --- a/todo.txt +++ b/todo.txt @@ -1,3 +1,5 @@ +* multiple knowledge bases, one for internal facts and one for each indexed paths +* perhaps a way to structure the prompt using <> tags. The memory items need to be distinct. * use poetry /* why is the cache not working? The system re-loads the knowledge every time diff --git a/wafl/readers/reader_factory.py b/wafl/readers/reader_factory.py index fbbbcd3b..14ccb70c 100644 --- a/wafl/readers/reader_factory.py +++ b/wafl/readers/reader_factory.py @@ -3,7 +3,7 @@ class ReaderFactory: - _chunk_size = 1000 + _chunk_size = 10000 _overlap = 100 _extension_to_reader_dict = {".pdf": PdfReader, ".txt": TextReader}