Lumia101/kanana-1.5-15.7b-a3b-instruct-Q4_K_M-GGUF

This model was converted to GGUF format from kakaocorp/kanana-1.5-15.7b-a3b-instruct using llama.cpp via the ggml.ai's GGUF-my-repo space. Refer to the original model card for more details on the model.

Use with llama.cpp

Install llama.cpp through brew (works on Mac and Linux)

brew install llama.cpp

Invoke the llama.cpp server or the CLI.

CLI:

llama-cli --hf-repo Lumia101/kanana-1.5-15.7b-a3b-instruct-Q4_K_M-GGUF --hf-file kanana-1.5-15.7b-a3b-instruct-q4_k_m.gguf -p "The meaning to life and the universe is"

Server:

llama-server --hf-repo Lumia101/kanana-1.5-15.7b-a3b-instruct-Q4_K_M-GGUF --hf-file kanana-1.5-15.7b-a3b-instruct-q4_k_m.gguf -c 2048

Note: You can also use this checkpoint directly through the usage steps listed in the Llama.cpp repo as well.

Step 1: Clone llama.cpp from GitHub.

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

Step 2: Move into the llama.cpp folder and build it with LLAMA_CURL=1 flag along with other hardware-specific flags (for ex: LLAMA_CUDA=1 for Nvidia GPUs on Linux).

cd llama.cpp && LLAMA_CURL=1 make

Step 3: Run inference through the main binary.

./llama-cli --hf-repo Lumia101/kanana-1.5-15.7b-a3b-instruct-Q4_K_M-GGUF --hf-file kanana-1.5-15.7b-a3b-instruct-q4_k_m.gguf -p "The meaning to life and the universe is"

or

./llama-server --hf-repo Lumia101/kanana-1.5-15.7b-a3b-instruct-Q4_K_M-GGUF --hf-file kanana-1.5-15.7b-a3b-instruct-q4_k_m.gguf -c 2048

How to use this Model in Colab

1. Press "Use this model" and select "Google Colab".

2. Download the library by running the following code.

# 1) Clean up previously installed libraries in case of anomaly
!pip uninstall -y llama-cpp-python

# 2) Install llama-cpp-python in GPU-enabled mode
!pip install llama-cpp-python --extra-index-url https://abetlen.github.io/llama-cpp-python/whl/cu122

# If the installation fails, check the CUDA version with the following command and reinstall with the appropriate CUDA version.
# !nvidia-smi

3. Copy the following code into Colab and execute it.

from llama_cpp import Llama

llm = Llama.from_pretrained(
  repo_id="Lumia101/kanana-1.5-15.7b-a3b-instruct-Q4_K_M-GGUF",
  filename="kanana-1.5-15.7b-a3b-instruct-q4_k_m.gguf",
  n_gpu_layers=-1,
  n_ctx=16384, # If you are getting errors due to insufficient VRAM, adjust this value.
  verbose=True
)

llm.create_chat_completion (
  messages = [
    {
      "role": "user",
      "content": "What is the capital of Korea?" # If you want to ask this model a different question, replace "What is the capital of Korea?" with the question you want to ask.
    }
  ]
)