Use Ollama with Chatbot-Ollama to use local LLM files.

Just go

• Download and Install ollama from ollama.com or ollama.ai
• Open http://127.0.0.1:11434/ to show [ollama is running]
• Use PowerShell to use ollama
  command: ollama list , to list all model
  command: ollama rm modelname , to del model
• Install docker manager and use it to install chatbox-ollama
• Open http://127.0.0.1:3000/
• Use already download model files
• Create a Modelfile as your_modelname, and config LLM path like below
  FROM c:\path\your_modelname.gguf
  SYSTEM ""
• Open powershell and goto your_modelname file of directory, run command like below
  ollama create your_modelname -f ./your_modelname
  ollama run your_modelname(>> /bye to exit)
• Open http://127.0.0.1:3000/zh , chat with llm

Stable Diffusion for Win & Anaconda

 Install Env AI

• Clone https://github.com/AUTOMATIC1111/stable-diffusion-webui.git
• Download model https://civitai.com/ and move to stable-diffusion-webui\models\Stable-diffusion
• Download v1-5-pruned-emaonly.ckpt from https://huggingface.co/runwayml/stable-diffusion-v1-5 (use less vram), and move to stable-diffusion-webui\models\Stable-diffusion
• Settings->Face restoration click GFPGAN
  
• Run webui-user.bat
• Open http://127.0.0.1:7860

Env for AI

 Install Python https://www.python.org/downloads/ (allow use admin and add path)

Install Anaconda https://www.anaconda.com/download

Install pytorch https://pytorch.org/get-started/locally/ find cuda version and dl pip3 install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu118

Install Git, Lfs

gpt4all-j and use streamlit

 modify on 2023/10/26

git clone https://github.com/ajvikram/streamlit-gpt.git

cd streamlit-gpt 

create model directory and  download https://gpt4all.io/models/ggml-gpt4all-j.bin in this directory

back to streamlit-gpt directory

run adconda

conda create -n gptj python=='3.11.5'

conda activate -n gptj

pip install -r requirement.txt

streamlit run main.py --server.port=8080 --server.address=0.0.0.0<-your ip

GPT4All test

 Local

• Website https://github.com/nomic-ai/gpt4all#readme
• Use conda on win, and git
• git clone https://github.com/nomic-ai/gpt4all.git
• down gpt4all-lora-quantized.bin from https://github.com/nomic-ai/gpt4all/issues/246
• move bin file to gpt4all-training/chat
• run exe to test

CHT GPT on Local with Win

Note

• Env
• install python
• install Anaconda3, or Miniconda
• conda update conda
• conda install pip3
• install Git, Git LFS
• Go and Choose https://pytorch.org/get-started/locally/ , Get a url for next
• Run Anaconda powershell mode
• pip3 install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu118
• Create Directory for GPT (ex:GPT)
• Langchain-Chatchat: git clone https://github.com/chatchat-space/Langchain-Chatchat.git
• chatglm2-6b: git clone https://huggingface.co/THUDM/chatglm2-6b
• m3e-base: git clone https://huggingface.co/moka-ai/m3e-base
• Langchain-Chatchat/configs
• *.py.example copyto *.py
• modify server_config.py
  DEFAULT_BIND_HOST = "your ip"
• modify model_config.py
  find: modify:embedding_model_dict = {
  "m3e-base": "/your_path/GPT/m3e-base",
  check: EMBEDDING_MODEL = "m3e-base"
  find: modify:llm_model_dict = { "chatglm2-6b": {
  "local_model_path": "/your_path/GPT/chatglm2-6b",
  check: modify:LLM_MODEL = "chatglm2-6b"
• install Langchain-Chatchat and Run (Anaconda powershell mode)
  cd Langchain-Chatchat
  conda create -n Chatchat python==3.10
  conda active Chatchat
  pip3 install -r requirements.txt
  python init_database.py --recreate-vs (for first)
• run Langchain-Chatchat
  python startup.py -a (Run All)[Input Email for first, and active service]
• Check
• Firewall pass python
• Firewall pass 8501/TCP
• INFO
  https://zhuanlan.zhihu.com/p/645807338?utm_id=0
  https://zhuanlan.zhihu.com/p/652530137
  https://zhuanlan.zhihu.com/p/617208490
  https://zhuanlan.zhihu.com/p/630049721?utm_id=0