Instructions to use cloudyu/Mixtral_7Bx4_MOE_DPO with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use cloudyu/Mixtral_7Bx4_MOE_DPO with Transformers:

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

pipe = pipeline("text-generation", model="cloudyu/Mixtral_7Bx4_MOE_DPO")

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("cloudyu/Mixtral_7Bx4_MOE_DPO")
model = AutoModelForCausalLM.from_pretrained("cloudyu/Mixtral_7Bx4_MOE_DPO")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use cloudyu/Mixtral_7Bx4_MOE_DPO with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "cloudyu/Mixtral_7Bx4_MOE_DPO"
# 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_7Bx4_MOE_DPO",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/cloudyu/Mixtral_7Bx4_MOE_DPO

SGLang

How to use cloudyu/Mixtral_7Bx4_MOE_DPO 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_7Bx4_MOE_DPO" \
    --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_7Bx4_MOE_DPO",
		"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_7Bx4_MOE_DPO" \
        --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_7Bx4_MOE_DPO",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use cloudyu/Mixtral_7Bx4_MOE_DPO with Docker Model Runner:
```
docker model run hf.co/cloudyu/Mixtral_7Bx4_MOE_DPO
```

Train after merging?

by adi-kmt - opened Feb 19, 2024

Discussion

adi-kmt

Feb 19, 2024

Other than adding a positive prompt, is it necessary to further finetune after merging to an moe?

cloudyu

Owner Feb 19, 2024

fine-tune can improve the score again if the dataset is great.
https://huggingface.co/cloudyu/Pluto_24B_DPO_200/blob/main/dpo-metrics.jpg

rookiemango

Mar 20, 2024

Thanks for sharing, i would like to know if have you done the sft before doing dpo, and if yes on which dataset? if not, then could you tell me which comparison training you are using during dpo training ? thanks again for you sharing

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