This is not a fully functional program. It is a placeholder to work with and develop libraries based on Ollama.
References and sources are scattered throughout the code.
All apologies for anyone or anybody who was looking for a program well defined and documented.
Compile Notes: "ollama-hpp" should be in the same directory as "ollama_chat"
To Compile: "cd" into build. Run "cmak.sh" to build Cmake Files (unnecessary?) (Run Once) Run "m.sh" subsequencially for any changes to build the executable.
Dependancies Notes:
sudo apt-get update
sudo apt-get install nlohmann-json3-dev
sudo apt-get install libssl-dev
sudo apt install libcurl4-openssl-dev
sudo apt install libboost-filesystem-dev
cpp-httplib
gh repo clone yhirose/cpp-httplib
https://github.com/ollama/ollama
https://github.com/jmont-dev/ollama-hpp
gh repo clone jmont-dev/ollama-hpp
https://github.com/BlackyDrum/chromadb-cpp
gh repo clone BlackyDrum/chromadb-cpp
ChromaDB Notes:
https://docs.trychroma.com/deployment/auth#static-api-token-authentication
BuOllama Build / Install, things to do. (https://github.com/ollama/ollama/blob/main/docs/development.md)
CUDA Toolkit Installer: (not sure it works correctly) (https://developer.nvidia.com/cuda-downloads?target_os=Linux&target_arch=x86_64&Distribution=WSL-Ubuntu&target_version=2.0&target_type=runfile_local)
wget https://developer.download.nvidia.com/compute/cuda/12.6.2/local_installers/cuda_12.6.2_560.35.03_linux.run
sudo sh cuda_12.6.2_560.35.03_linux.run
verify -lcuda installed correctly. (If error, gives you pointer to the correct spot.)
nvcc --version
which may be
sudo apt install nvidia-cuda-toolkit
Python Notes:
sudo apt install python3-pip
sudo apt install python3.12-venv
python3 -m venv ./venv
source venv/bin/activate
pip install chromadb
pip list --outdated
pip3 install --upgrade package
pip3 install nltk
alternative instructions:
Update package list
sudo apt update
Install CUDA Toolkit (latest version)
sudo apt install -y cuda-toolkit-12-9
Verify CUDA installation
nvcc --version
Add CUDA to system PATH if needed
echo 'export PATH=/usr/local/cuda/bin:$PATH' >> ~/.bashrc
source ~/.bashrc
Install extra NVIDIA libraries (optional)
sudo apt install -y nvidia-cuda-toolkit nvidia-cudnn
Check GPU status
nvidia-smi
pip install scikit-learn
https://scikit-learn.org/stable/modules/generated/sklearn.metrics.pairwise_distances.html?form=MG0AV3
export OLLAMA_NUM_PARALLEL=3
export OLLAMA_MAX_LOADED_MODELS=3
Model manager Page doesnt load on my system https://smcleod.net/2024/06/gollama-ollama-model-manager/