Instructions to use cobrakenji/granite-34b-code-instruct-Q4_K_M-GGUF with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use cobrakenji/granite-34b-code-instruct-Q4_K_M-GGUF with Transformers:

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

pipe = pipeline("text-generation", model="cobrakenji/granite-34b-code-instruct-Q4_K_M-GGUF")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoModel
model = AutoModel.from_pretrained("cobrakenji/granite-34b-code-instruct-Q4_K_M-GGUF", dtype="auto")

llama-cpp-python

How to use cobrakenji/granite-34b-code-instruct-Q4_K_M-GGUF with llama-cpp-python:

# !pip install llama-cpp-python

from llama_cpp import Llama

llm = Llama.from_pretrained(
	repo_id="cobrakenji/granite-34b-code-instruct-Q4_K_M-GGUF",
	filename="granite-34b-code-instruct-q4_k_m.gguf",
)

llm.create_chat_completion(
	messages = [
		{
			"role": "user",
			"content": "What is the capital of France?"
		}
	]
)

Notebooks
Google Colab
Kaggle
Local Apps

llama.cpp

How to use cobrakenji/granite-34b-code-instruct-Q4_K_M-GGUF with llama.cpp:

Install from brew

brew install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama-server -hf cobrakenji/granite-34b-code-instruct-Q4_K_M-GGUF:Q4_K_M
# Run inference directly in the terminal:
llama-cli -hf cobrakenji/granite-34b-code-instruct-Q4_K_M-GGUF:Q4_K_M

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama-server -hf cobrakenji/granite-34b-code-instruct-Q4_K_M-GGUF:Q4_K_M
# Run inference directly in the terminal:
llama-cli -hf cobrakenji/granite-34b-code-instruct-Q4_K_M-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-34b-code-instruct-Q4_K_M-GGUF:Q4_K_M
# Run inference directly in the terminal:
./llama-cli -hf cobrakenji/granite-34b-code-instruct-Q4_K_M-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-34b-code-instruct-Q4_K_M-GGUF:Q4_K_M
# Run inference directly in the terminal:
./build/bin/llama-cli -hf cobrakenji/granite-34b-code-instruct-Q4_K_M-GGUF:Q4_K_M

Use Docker

docker model run hf.co/cobrakenji/granite-34b-code-instruct-Q4_K_M-GGUF:Q4_K_M

LM Studio
Jan

vLLM

How to use cobrakenji/granite-34b-code-instruct-Q4_K_M-GGUF with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "cobrakenji/granite-34b-code-instruct-Q4_K_M-GGUF"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "cobrakenji/granite-34b-code-instruct-Q4_K_M-GGUF",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/cobrakenji/granite-34b-code-instruct-Q4_K_M-GGUF:Q4_K_M

SGLang

How to use cobrakenji/granite-34b-code-instruct-Q4_K_M-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-34b-code-instruct-Q4_K_M-GGUF" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "cobrakenji/granite-34b-code-instruct-Q4_K_M-GGUF",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

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-34b-code-instruct-Q4_K_M-GGUF" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "cobrakenji/granite-34b-code-instruct-Q4_K_M-GGUF",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Ollama
How to use cobrakenji/granite-34b-code-instruct-Q4_K_M-GGUF with Ollama:
```
ollama run hf.co/cobrakenji/granite-34b-code-instruct-Q4_K_M-GGUF:Q4_K_M
```

Unsloth Studio new

How to use cobrakenji/granite-34b-code-instruct-Q4_K_M-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-34b-code-instruct-Q4_K_M-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-34b-code-instruct-Q4_K_M-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-34b-code-instruct-Q4_K_M-GGUF to start chatting

Docker Model Runner
How to use cobrakenji/granite-34b-code-instruct-Q4_K_M-GGUF with Docker Model Runner:
```
docker model run hf.co/cobrakenji/granite-34b-code-instruct-Q4_K_M-GGUF:Q4_K_M
```

Lemonade

How to use cobrakenji/granite-34b-code-instruct-Q4_K_M-GGUF with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull cobrakenji/granite-34b-code-instruct-Q4_K_M-GGUF:Q4_K_M

Run and chat with the model

lemonade run user.granite-34b-code-instruct-Q4_K_M-GGUF-Q4_K_M

List all available models

lemonade list

cobrakenji/granite-34b-code-instruct-Q4_K_M-GGUF

This model was converted to GGUF format from ibm-granite/granite-34b-code-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 --hf-repo cobrakenji/granite-34b-code-instruct-Q4_K_M-GGUF --hf-file granite-34b-code-instruct-q4_k_m.gguf -p "The meaning to life and the universe is"

Server:

llama-server --hf-repo cobrakenji/granite-34b-code-instruct-Q4_K_M-GGUF --hf-file granite-34b-code-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.

./main --hf-repo cobrakenji/granite-34b-code-instruct-Q4_K_M-GGUF --hf-file granite-34b-code-instruct-q4_k_m.gguf -p "The meaning to life and the universe is"

./server --hf-repo cobrakenji/granite-34b-code-instruct-Q4_K_M-GGUF --hf-file granite-34b-code-instruct-q4_k_m.gguf -c 2048

Downloads last month: 32

GGUF

Model size

34B params

Architecture

starcoder

Hardware compatibility

4-bit

Model tree for cobrakenji/granite-34b-code-instruct-Q4_K_M-GGUF

Base model

ibm-granite/granite-34b-code-base-8k

Finetuned

ibm-granite/granite-34b-code-instruct-8k

Quantized

(11)

this model

Datasets used to train cobrakenji/granite-34b-code-instruct-Q4_K_M-GGUF

Evaluation results

pass@1 on HumanEvalSynthesis(Python)
self-reported

62.200
pass@1 on HumanEvalSynthesis(Python)
self-reported

56.700
pass@1 on HumanEvalSynthesis(Python)
self-reported

62.800
pass@1 on HumanEvalSynthesis(Python)
self-reported

47.600
pass@1 on HumanEvalSynthesis(Python)
self-reported

57.900
pass@1 on HumanEvalSynthesis(Python)
self-reported

41.500
pass@1 on HumanEvalSynthesis(Python)
self-reported

53.000
pass@1 on HumanEvalSynthesis(Python)
self-reported

45.100
pass@1 on HumanEvalSynthesis(Python)
self-reported

50.600
pass@1 on HumanEvalSynthesis(Python)
self-reported

36.000
pass@1 on HumanEvalSynthesis(Python)
self-reported

42.700
pass@1 on HumanEvalSynthesis(Python)
self-reported

23.800
pass@1 on HumanEvalSynthesis(Python)
self-reported

54.900
pass@1 on HumanEvalSynthesis(Python)
self-reported

47.600
pass@1 on HumanEvalSynthesis(Python)
self-reported

55.500
pass@1 on HumanEvalSynthesis(Python)
self-reported

51.200