Instructions to use nebuxcloud/Falcon3-10B-Instruct-1.58bit-GGUF with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use nebuxcloud/Falcon3-10B-Instruct-1.58bit-GGUF with Transformers:

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

pipe = pipeline("text-generation", model="nebuxcloud/Falcon3-10B-Instruct-1.58bit-GGUF")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("nebuxcloud/Falcon3-10B-Instruct-1.58bit-GGUF")
model = AutoModelForCausalLM.from_pretrained("nebuxcloud/Falcon3-10B-Instruct-1.58bit-GGUF")
messages = [
    {"role": "user", "content": "Who are you?"},
]
inputs = tokenizer.apply_chat_template(
	messages,
	add_generation_prompt=True,
	tokenize=True,
	return_dict=True,
	return_tensors="pt",
).to(model.device)

outputs = model.generate(**inputs, max_new_tokens=40)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use nebuxcloud/Falcon3-10B-Instruct-1.58bit-GGUF with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "nebuxcloud/Falcon3-10B-Instruct-1.58bit-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": "nebuxcloud/Falcon3-10B-Instruct-1.58bit-GGUF",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/nebuxcloud/Falcon3-10B-Instruct-1.58bit-GGUF

SGLang

How to use nebuxcloud/Falcon3-10B-Instruct-1.58bit-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 "nebuxcloud/Falcon3-10B-Instruct-1.58bit-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": "nebuxcloud/Falcon3-10B-Instruct-1.58bit-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 "nebuxcloud/Falcon3-10B-Instruct-1.58bit-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": "nebuxcloud/Falcon3-10B-Instruct-1.58bit-GGUF",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use nebuxcloud/Falcon3-10B-Instruct-1.58bit-GGUF with Docker Model Runner:
```
docker model run hf.co/nebuxcloud/Falcon3-10B-Instruct-1.58bit-GGUF
```

Model Card for Falcon3-10B-Instruct-1.58bit-q2b0

Falcon3-10B-1.58 Models

The Falcon3-10B-1.58bit-q2b0 is a quantized version of Falcon3-10B-Instruct, leveraging the q2b0 quantization method from Candle. This enables extreme compression while maintaining strong performance across various NLP tasks.

Model Details

Model Sources

Repository: tiiuae/Falcon3-10B-Instruct
Quantization PR: Candle q2b0 Quantization

Quantization Details

The model has been quantized using the q2b0 method from Candle. This approach reduces model size significantly while preserving performance. For more details on this quantization technique, refer to the Candle PR #2683.