Instructions to use moonshotai/Moonlight-16B-A3B with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use moonshotai/Moonlight-16B-A3B with Transformers:

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

pipe = pipeline("text-generation", model="moonshotai/Moonlight-16B-A3B", trust_remote_code=True)
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("moonshotai/Moonlight-16B-A3B", trust_remote_code=True)
model = AutoModelForCausalLM.from_pretrained("moonshotai/Moonlight-16B-A3B", trust_remote_code=True)
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 Settings

vLLM

How to use moonshotai/Moonlight-16B-A3B with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "moonshotai/Moonlight-16B-A3B"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "moonshotai/Moonlight-16B-A3B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/moonshotai/Moonlight-16B-A3B

SGLang

How to use moonshotai/Moonlight-16B-A3B 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 "moonshotai/Moonlight-16B-A3B" \
    --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": "moonshotai/Moonlight-16B-A3B",
		"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 "moonshotai/Moonlight-16B-A3B" \
        --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": "moonshotai/Moonlight-16B-A3B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use moonshotai/Moonlight-16B-A3B with Docker Model Runner:
```
docker model run hf.co/moonshotai/Moonlight-16B-A3B
```

Moonlight-16B-A3B training fails in DeepseekV3MoE.forward with UnboundLocalError when shared experts are enabled

by arindamm - opened Mar 14

Discussion

arindamm

Mar 14

Root Cause
In DeepseekV3MoE.forward, the variable y was assigned only in the inference branch:

if not self.training:
y = self.moe_infer(hidden_states, topk_idx, topk_weight).view(*orig_shape)
if self.config.n_shared_experts is not None:
y = y + self.shared_experts(identity)
return y
During training, execution skipped the if not self.training branch, then attempted to use y in the shared-expert accumulation step.

That caused:

UnboundLocalError: cannot access local variable 'y' where it is not associated with a value
Impact
Inference can proceed.
Fine-tuning/training fails on the first forward pass.
This is an upstream model-code bug, not a dataset or trainer configuration issue.
Local Fix Applied
Patched file:

.cache/huggingface/modules/transformers_modules/moonshotai/Moonlight_hyphen_16B_hyphen_A3B/476b36a473d4467f94469414bef6cee75c9c8172/modeling_deepseek.py
Fix applied:

Added a training-mode routed-expert computation path in DeepseekV3MoE.forward.
Ensured y is always initialized before shared experts are added.
Kept the inference path using moe_infer(...) unchanged.

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