Instructions to use cobrakenji/granite-20b-code-base-GGUF with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use cobrakenji/granite-20b-code-base-GGUF with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("text-generation", model="cobrakenji/granite-20b-code-base-GGUF")

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("cobrakenji/granite-20b-code-base-GGUF")
model = AutoModelForCausalLM.from_pretrained("cobrakenji/granite-20b-code-base-GGUF", device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

llama.cpp

How to use cobrakenji/granite-20b-code-base-GGUF with llama.cpp:

Install (macOS, Linux)

curl -LsSf https://llama.app/install.sh | sh
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf cobrakenji/granite-20b-code-base-GGUF:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf cobrakenji/granite-20b-code-base-GGUF:Q4_K_M

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf cobrakenji/granite-20b-code-base-GGUF:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf cobrakenji/granite-20b-code-base-GGUF:Q4_K_M

Use pre-built binary

# Download pre-built binary from:
# https://github.com/ggerganov/llama.cpp/releases
# Start a local OpenAI-compatible server with a web UI:
./llama-server -hf cobrakenji/granite-20b-code-base-GGUF:Q4_K_M
# Run inference directly in the terminal:
./llama-cli -hf cobrakenji/granite-20b-code-base-GGUF:Q4_K_M

Build from source code

git clone https://github.com/ggerganov/llama.cpp.git
cd llama.cpp
cmake -B build
cmake --build build -j --target llama-server llama-cli
# Start a local OpenAI-compatible server with a web UI:
./build/bin/llama-server -hf cobrakenji/granite-20b-code-base-GGUF:Q4_K_M
# Run inference directly in the terminal:
./build/bin/llama-cli -hf cobrakenji/granite-20b-code-base-GGUF:Q4_K_M

Use Docker

docker model run hf.co/cobrakenji/granite-20b-code-base-GGUF:Q4_K_M

LM Studio
Jan

vLLM

How to use cobrakenji/granite-20b-code-base-GGUF with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "cobrakenji/granite-20b-code-base-GGUF"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "cobrakenji/granite-20b-code-base-GGUF",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/cobrakenji/granite-20b-code-base-GGUF:Q4_K_M

SGLang

How to use cobrakenji/granite-20b-code-base-GGUF with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "cobrakenji/granite-20b-code-base-GGUF" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "cobrakenji/granite-20b-code-base-GGUF",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker images

docker run --gpus all \
    --shm-size 32g \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server \
        --model-path "cobrakenji/granite-20b-code-base-GGUF" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "cobrakenji/granite-20b-code-base-GGUF",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Ollama
How to use cobrakenji/granite-20b-code-base-GGUF with Ollama:
```
ollama run hf.co/cobrakenji/granite-20b-code-base-GGUF:Q4_K_M
```

Unsloth Studio

How to use cobrakenji/granite-20b-code-base-GGUF with Unsloth Studio:

Install Unsloth Studio (macOS, Linux, WSL)

curl -fsSL https://unsloth.ai/install.sh | sh
# Run unsloth studio
unsloth studio -H 0.0.0.0 -p 8888
# Then open http://localhost:8888 in your browser
# Search for cobrakenji/granite-20b-code-base-GGUF to start chatting

Install Unsloth Studio (Windows)

irm https://unsloth.ai/install.ps1 | iex
# Run unsloth studio
unsloth studio -H 0.0.0.0 -p 8888
# Then open http://localhost:8888 in your browser
# Search for cobrakenji/granite-20b-code-base-GGUF to start chatting

Using HuggingFace Spaces for Unsloth

# No setup required
# Open https://huggingface.co/spaces/unsloth/studio in your browser
# Search for cobrakenji/granite-20b-code-base-GGUF to start chatting

Docker Model Runner
How to use cobrakenji/granite-20b-code-base-GGUF with Docker Model Runner:
```
docker model run hf.co/cobrakenji/granite-20b-code-base-GGUF:Q4_K_M
```

Lemonade

How to use cobrakenji/granite-20b-code-base-GGUF with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull cobrakenji/granite-20b-code-base-GGUF:Q4_K_M

Run and chat with the model

lemonade run user.granite-20b-code-base-GGUF-Q4_K_M

List all available models

lemonade list

Atomic Chat

Description:

This is forked from IBM's granite-20b-code-base-GGUF - commit d70433a71e2fb9e20f8bfca3ff2d8c15393f0e44.
Refer to the original model card for more details.

Use with llama.cpp

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

# install
make

# run generation
./main -m granite-20b-code-base-GGUF/granite-20b-code-base.Q4_K_M.gguf -n 128 -p "def generate_random(x: int):" --color

Downloads last month: 48

GGUF

Model size

20B params

Architecture

starcoder

Hardware compatibility

4-bit

Datasets used to train cobrakenji/granite-20b-code-base-GGUF

Evaluation results

pass@1 on MBPP
self-reported

43.800
pass@1 on MBPP+
self-reported

51.600
pass@1 on HumanEvalSynthesis(Python)
self-reported

48.200
pass@1 on HumanEvalSynthesis(JavaScript)
self-reported

50.000
pass@1 on HumanEvalSynthesis(Java)
self-reported

59.100
pass@1 on HumanEvalSynthesis(Go)
self-reported

32.300
pass@1 on HumanEvalSynthesis(C++)
self-reported

40.900
pass@1 on HumanEvalSynthesis(Rust)
self-reported

35.400