Instructions to use cloudyu/Mixtral_34Bx2_MoE_60B with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use cloudyu/Mixtral_34Bx2_MoE_60B with Transformers:

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

pipe = pipeline("text-generation", model="cloudyu/Mixtral_34Bx2_MoE_60B")

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("cloudyu/Mixtral_34Bx2_MoE_60B")
model = AutoModelForCausalLM.from_pretrained("cloudyu/Mixtral_34Bx2_MoE_60B")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use cloudyu/Mixtral_34Bx2_MoE_60B with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "cloudyu/Mixtral_34Bx2_MoE_60B"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "cloudyu/Mixtral_34Bx2_MoE_60B",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/cloudyu/Mixtral_34Bx2_MoE_60B

SGLang

How to use cloudyu/Mixtral_34Bx2_MoE_60B 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 "cloudyu/Mixtral_34Bx2_MoE_60B" \
    --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": "cloudyu/Mixtral_34Bx2_MoE_60B",
		"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 "cloudyu/Mixtral_34Bx2_MoE_60B" \
        --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": "cloudyu/Mixtral_34Bx2_MoE_60B",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use cloudyu/Mixtral_34Bx2_MoE_60B with Docker Model Runner:
```
docker model run hf.co/cloudyu/Mixtral_34Bx2_MoE_60B
```

The function_calling and translation abilities are weaker than Mixtral 8x7b

#11

by bingw5 - opened Jan 16, 2024

Discussion

bingw5

Jan 16, 2024

As per my test, this model surpassed mixtral 8x7b generally, especially in math and inference. However, it's also weaker than Mixtral while dealing with function calling. I used the same request body, and expect the model to generate proper json response represent a function call. This model responses "I am not able to access internet...", while Mixtral 8x7b success every time. Translation from Chinese to English seems slightly weaker than Mixtral 8x7b, this model failed to translate some terms or rare words.
In conclusion, this model is really amazing for most of the scenarios, you really did a great job! I hope the new versions can enhance these two weakness, make it better.

cloudyu

Owner Jan 16, 2024

Many thanks for your feedback.

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