Instructions to use QuantPanda/granite-uncertainty-3.2-8b-lora-GGUF with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use QuantPanda/granite-uncertainty-3.2-8b-lora-GGUF with Transformers:

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

pipe = pipeline("text-generation", model="QuantPanda/granite-uncertainty-3.2-8b-lora-GGUF")

# Load model directly
from transformers import AutoModel
model = AutoModel.from_pretrained("QuantPanda/granite-uncertainty-3.2-8b-lora-GGUF", device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use QuantPanda/granite-uncertainty-3.2-8b-lora-GGUF with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "QuantPanda/granite-uncertainty-3.2-8b-lora-GGUF"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "QuantPanda/granite-uncertainty-3.2-8b-lora-GGUF",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/QuantPanda/granite-uncertainty-3.2-8b-lora-GGUF

SGLang

How to use QuantPanda/granite-uncertainty-3.2-8b-lora-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 "QuantPanda/granite-uncertainty-3.2-8b-lora-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": "QuantPanda/granite-uncertainty-3.2-8b-lora-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 "QuantPanda/granite-uncertainty-3.2-8b-lora-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": "QuantPanda/granite-uncertainty-3.2-8b-lora-GGUF",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use QuantPanda/granite-uncertainty-3.2-8b-lora-GGUF with Docker Model Runner:
```
docker model run hf.co/QuantPanda/granite-uncertainty-3.2-8b-lora-GGUF
```

GGUF Quants for the lora adapter: ibm-granite/granite-uncertainty-3.2-8b-lora

Link to the original repo: https://huggingface.co/ibm-granite/granite-uncertainty-3.2-8b-lora

You need the instruct gguf to apply this lora to (ie. granite-3.2-8B-instruct-Q4_K_M.gguf)

Then run it like this:

llama-cli -m granite-3.2-8B-instruct-Q4_K_M.gguf --lora granite-uncertainty-3.2-8b-lora-f16.gguf --conversation --jinja

To get the certainty score, simply paste this line into the chat, after getting the first reply:

<|end_of_role|>\n<|start_of_role|>certainty<|end_of_role|>

It's a bit hacky, but it works for now.

Example of what it should look like:

Downloads last month: 27

GGUF

Model size

2.95M params

Architecture

granite

Hardware compatibility

8-bit

16-bit

32-bit

Model tree for QuantPanda/granite-uncertainty-3.2-8b-lora-GGUF

Base model

ibm-granite/granite-3.2-8b-lora-uncertainty

Quantized

(1)

this model