Skip to content

Latest commit

 

History

History
21 lines (15 loc) · 726 Bytes

File metadata and controls

21 lines (15 loc) · 726 Bytes

Setup

create dev environment:

  • conda create poke-llm
  • conda activate poke-llm
  • pip install -r requirements.txt

View Models

  • ollama list

Create Models with modelfile:

  • ollama create model-name -f Modelfile

Finetuneing Model:

  1. use finetune_model.py to perform LoRA finetune with model on .JSON dataset. Script will output a lora file
  2. Merge the finetuned lora file with the base model to create the finetuned model. This will output a merged folder
  3. Navigate to llama.cpp and use the script 'convert_hf_to_gguf_update.py' with command - python3 convert_hf_to_gguf_update.py --outtype f16 --outfile ./your-model.gguf ./merged-model

Run Pokemon Chatbot

  • python pokemon-chatbot.py in /scripts