Instructions to use manycore-research/SpatialLM-Llama-1B with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use manycore-research/SpatialLM-Llama-1B with Transformers:

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

pipe = pipeline("text-generation", model="manycore-research/SpatialLM-Llama-1B")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoModelForCausalLM
model = AutoModelForCausalLM.from_pretrained("manycore-research/SpatialLM-Llama-1B", dtype="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use manycore-research/SpatialLM-Llama-1B with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "manycore-research/SpatialLM-Llama-1B"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "manycore-research/SpatialLM-Llama-1B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/manycore-research/SpatialLM-Llama-1B

SGLang

How to use manycore-research/SpatialLM-Llama-1B 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 "manycore-research/SpatialLM-Llama-1B" \
    --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": "manycore-research/SpatialLM-Llama-1B",
		"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 "manycore-research/SpatialLM-Llama-1B" \
        --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": "manycore-research/SpatialLM-Llama-1B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use manycore-research/SpatialLM-Llama-1B with Docker Model Runner:
```
docker model run hf.co/manycore-research/SpatialLM-Llama-1B
```

Local Installation Video and Testing - Step by Step

by fahdmirzac - opened Mar 21, 2025

Discussion

fahdmirzac

Mar 21, 2025

Hi,
Kudos on producing such a sublime model. I did a local installation and testing video :

https://youtu.be/dy20K-6Pt4U?si=BZdp5RaofMpNI2Lj

Thanks and regards,
Fahd

LeeSandis

Mar 23, 2025

I wanted to test Qwen version on my old 1080 NVidia GPU with 8 GB is that at all possible?
Also I followed the instructions to the letter.

But having trouble installing was nvidia-nccl-cu12. It says it can't be found. Tried a lot of different suggestions to fix that. Can this one be skipped? , how. Or found?

"NCCL stands for NVIDIA Collective Communications Library, which is primarily used for multi-GPU communications.
NCCL is a library that enables efficient communication (message passing) between multiple GPUs, both within and across multiple nodes. It's particularly important for distributed deep learning training across multiple GPUs.
The "cu12" part in the name indicates it's for CUDA 12.x compatibility."

ykapch2

Mar 23, 2025

Upload images, audio, and videos by dragging in the text input, pasting, or clicking here.

Tap or paste here to upload images

· Sign up or log in to comment