Instructions to use mrfakename/EmoAct-MiMo with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use mrfakename/EmoAct-MiMo with PEFT:

from peft import PeftModel
from transformers import AutoModelForCausalLM

base_model = AutoModelForCausalLM.from_pretrained("XiaomiMiMo/MiMo-Audio-7B-Instruct")
model = PeftModel.from_pretrained(base_model, "mrfakename/EmoAct-MiMo")

Transformers

How to use mrfakename/EmoAct-MiMo with Transformers:

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

pipe = pipeline("text-generation", model="mrfakename/EmoAct-MiMo")

# Load model directly
from transformers import AutoModel
model = AutoModel.from_pretrained("mrfakename/EmoAct-MiMo", dtype="auto")

Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use mrfakename/EmoAct-MiMo with vLLM:

Install from pip and serve model

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

Use Docker

docker model run hf.co/mrfakename/EmoAct-MiMo

SGLang

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

Docker Model Runner
How to use mrfakename/EmoAct-MiMo with Docker Model Runner:
```
docker model run hf.co/mrfakename/EmoAct-MiMo
```

A newer version of this model is available: mrfakename/EmoAct-MiMo-v1.2

You need to agree to share your contact information to access this model

This repository is publicly accessible, but you have to accept the conditions to access its files and content.

Huge thanks to Hugging Face for supporting the compute for this model!

Audio samples (thanks to Christoph from LAION for generating them):

Very early checkpoint, but it's a good model/finetune.

Base model: MiMo Audio

Prompt format:

Emotion: <emotion>
Text: <text>

Example:

Emotion: intense anger, rage, fury, hatred, and annoyance, speaking without any accent
Text: You know what? I'm done. I'm done with your excuses. (sharp exhale) Every single time, it's the same, and I actually believed you'd change. (voice cracks slightly) God, I'm such an idiot for trusting you again.

Training code (private): https://github.com/fakerybakery/mimo2

Voice cloning is not supported yet, coming soon.

Questions/feedback: DM realmrfakename on discord