Llama-2-7b-ft-instruct-es-GGUF

This project aims to convert clibrain/Llama-2-7b-ft-instruct-es model into a compatible llama.cpp GGUF model.

Content

1. Setup
2. Contributing

---

Setup

1. Clone llama.cpp

   git clone https://github.com/ggerganov/llama.cpp.git

2. Build llama.cpp

   cd llama.cpp && \
     make

3. Download test model

   bash download_test_model.sh

4. Test build

   llama.cpp/main -ins \
      -f llama.cpp/prompts/alpaca.txt \
      -t 8 \
      -ngl 1 \
      -m models/test/llama-2-7b.Q4_K_M.gguf \
      -c 2048 \
      -s 42 \
      -n -1 \
      --temp 0.7 \
      --repeat-penalty 1.1 \
      --color

5. Write a sample instruction: write a haiku about programming

Results:

Getting Started

To download the model clibrain/Llama-2-7b-ft-instruct-es, run:

python scripts/download_hf_model.py clibrain/Llama-2-7b-ft-instruct-es

It should take around 20min to download (based on your internet speed)

After, you can check files with:

ls -lhia models/Llama-2-7b-ft-instruct-es/

Contributing

Installing pre-commit

Pre-commit is already part of this project dependencies. If you would like to installed it as standalone run:

pip install pre-commit

To activate pre-commit run the following commands:

• Install Git hooks:

pre-commit install

• Update current hooks:

pre-commit autoupdate

To test your installation of pre-commit run:

pre-commit run --all-files