Instructions to use microsoft/GRIN-MoE with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use microsoft/GRIN-MoE with Transformers:

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

pipe = pipeline("text-generation", model="microsoft/GRIN-MoE", trust_remote_code=True)
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import GRIN-MoE
model = GRIN-MoE.from_pretrained("microsoft/GRIN-MoE", trust_remote_code=True, dtype="auto")

Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use microsoft/GRIN-MoE with vLLM:

Install from pip and serve model

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

Use Docker

docker model run hf.co/microsoft/GRIN-MoE

SGLang

How to use microsoft/GRIN-MoE 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 "microsoft/GRIN-MoE" \
    --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": "microsoft/GRIN-MoE",
		"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 "microsoft/GRIN-MoE" \
        --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": "microsoft/GRIN-MoE",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use microsoft/GRIN-MoE with Docker Model Runner:
```
docker model run hf.co/microsoft/GRIN-MoE
```

Question about sparsemixer

by DhanOS - opened Sep 25, 2024

Discussion

DhanOS

Sep 25, 2024

•

edited Sep 25, 2024

Hi there!
First of all - great work! :)

I'm experimenting with sparsemixer on a different model architecture and I'm looking at the sparsemixer's code for 2 reasons:

to make it work with DeepSpeed (DeepSpeed hangs after a few steps during code testing)
to make it work with top_k>2

I have 3 questions if you don't mind:

So, there are 2 almost identical blocks of code in sparsemixer that could be put in a loop, I guess. The only difference is line 834 (the first block, for the first expert):

            torch.rand_like(max_scores) > 0.75 # Heun's third-order method: f(x) - f(0) = .25 f'(x) + .75 f'(x/3.)

and line 881 (for the second expert):

            torch.rand_like(max_scores).uniform_() > 0.75 # Heun's third-order method: f(x) - f(0) = .25 f'(x) + .75 f'(x/3.)

is this draw from a uniform distribution added there for any particular reason? It's drawing two times now - once during rand_like call and then during uniform_ call.

Have you tested it with DeepSpeed ZeRO3? It may be an issue with how I implemented sparsemixer into my experiment, but the same model with softmax+topk works (trains) just fine (the modeling code contains a workaround for hangs with ZeRO3, I understand what the problem is).
Are there any additional considerations to make it work with top_k>2, or is top_k=2 just implemented this way for this experiment with the model you trained?

Thank you

LiyuanLucasLiu

Sep 25, 2024

there is no particular reason and the second .uniform_() could be removed;
Yes, we tried that. We ends up using ZeRO1 + PP + activation checkpointing, which yields the best throughput (much better than ZeRO3);
Yes, there are:

we model the sampling of the top-k as iterative sampling, which brings issues. As an example, if k is 4, and the four expert you get is a,b,c,d. Sampling a->b->c->d and d->c->b->a yield the same set of experts, but are treated separately. This causes complications on gradient computation
when there are multiple activated experts, it is a little bit tricky to integrate the first-order and third-order estimator. To start, I would recommond you to use the 1st-order estimator only (use set mask_for_one to 1.0 in all cases)

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